archive/blender-archive
Archived
1
1
Fork 0
Code Pull Requests Packages Activity
This repository has been archived on 2023-10-09. You can view files and clone it, but cannot push or open issues or pull requests.
Files
5d3e07862cceb7dc356e30b061dec065e57a7d03
blender-archive/source/blender/gpu/intern/gpu_extensions.c

2355 lines
62 KiB
C
Raw Normal View History

doxygen: prevent GPL license block from being parsed as doxygen comment.
2011-02-23 10:52:22 +00:00
/*
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
* ***** BEGIN GPL LICENSE BLOCK *****
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version 2
remove header text: "The Blender Foundation also sells licenses for use in proprietary software under the Blender Licens" also remove NaN references from files that have been added since blender went opensource.
2011-11-29 10:54:47 +00:00
* of the License, or (at your option) any later version.
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software Foundation,
correct fsf address
2010-02-12 13:34:04 +00:00
* Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
*
* The Original Code is Copyright (C) 2005 Blender Foundation.
* All rights reserved.
*
* The Original Code is: all of this file.
*
* Contributor(s): Brecht Van Lommel.
*
* ***** END GPL LICENSE BLOCK *****
*/
doxygen: blender/gpu tagged.
2011-02-27 20:25:53 +00:00
/** \file blender/gpu/intern/gpu_extensions.c
* \ingroup gpu
Docs: doxygen file descriptions for BLF, GPU and WM
2014-01-19 23:14:24 +11:00
*
* Wrap OpenGL features such as textures, shaders and GLSL
* with checks for drivers and GPU support.
doxygen: blender/gpu tagged.
2011-02-27 20:25:53 +00:00
*/
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
#include "DNA_image_types.h"
#include "MEM_guardedalloc.h"
#include "BLI_blenlib.h"
split BKE_utildefines.h, now it only has blender specific defines like GS() MAKE_ID, FILE_MAXDIR, moved the generic defines to BLI_utildefines.h. no functional changes.
2011-01-07 18:36:47 +00:00
#include "BLI_utildefines.h"
static functions for getting power of 2 values were being copied about too much, add to the BLI_math api. - is_power_of_2_i - power_of_2_min_i - power_of_2_max_i
2011-12-16 09:25:07 +00:00
#include "BLI_math_base.h"
Potential fix for T43987, ambient occlusion different between offscreen and on screen rendering. Aaaaah, the beauty of driver implementations of OpenGL! Turns out the problem here is that drivers calculate df/dy differently in some cases (probably because OpenGL counts y reverse to how the window system does, so drivers can get confused). Fixed this for the ATI case based on info we have so far, there's also the Intel case which will be handled separately (missing info on Intel's renderer string etc). Unfortunately we can't really fix this for the general case so we'll have to haldle cases as they come in our tracker and by adding silly string comparisons in our GPU initialization module <sigh>.
2015-03-30 14:14:32 +02:00
#include "BLI_math_vector.h"
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
header cleanup, include BLI before BKE, also use bool for ntreeShaderExecTree
2013-01-24 21:57:13 +00:00
#include "BKE_global.h"
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
#include "GPU_glew.h"
#include "GPU_debug.h"
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
#include "GPU_draw.h"
#include "GPU_extensions.h"
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
#include "GPU_compositing.h"
OpenGL: more work on fixed function lighting implementation as GLSL. * Rename functions and move to own header. * Add wrapper functions for glLight. * Auto detect if we can use faster code for solid lighting. * Various fixes for textured draw mode.
2013-03-10 15:38:23 +00:00
#include "GPU_simple_shader.h"
Ghost Context Refactor https://developer.blender.org/D643 Separates graphics context creation from window code in Ghost so that they can vary separately.
2014-10-07 15:46:19 -05:00
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
#include "intern/gpu_private.h"
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
code cleanup: dont include BLI_winstuff.h on non windows systems, also cleanup some defines/includes for windows.
2012-04-15 07:54:07 +00:00
#ifdef WIN32
# include "BLI_winstuff.h"
#endif
Ensure gpu_extensions.c compiles with MSVC -> include BLI_winstuff.h
2011-03-07 00:49:21 +00:00
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
/* TODO(sergey): Find better default values for this constants. */
#define MAX_DEFINE_LENGTH 1024
#define MAX_EXT_DEFINE_LENGTH 1024
Another attempt for T40981, clipping border does not work with GLSL on ATIs. This is actually a test to see if this can be enabled on ATI cards. According to various sources, newer ATI cards supporting GLSL 3.0 support gl_ClippingDistance in shaders, which is the forward compatible way to do custom clipping. This fix will bind 6 additional varying variables on ATIs, which may lead to some shaders not compiling due to limiting out of those variables, or to performance degradation. Also I do not have an ATI handy to test. Having those in mind, this commit may well be reverted later. Clipping planes are usually 4 (6 is for cube clipping), but making shaders depend on viewport state is really bad, and would lead to recompilation, so I took the worst case here to avoid that. Hopefully driver does some optimization there.
2014-07-11 19:17:29 +03:00
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
/* Extensions support */
/* extensions used:
style cleanup: comment blocks
2012-03-09 18:28:30 +00:00
* - texture border clamp: 1.3 core
style cleanup: correct typos
2012-04-21 14:14:58 +00:00
* - fragment shader: 2.0 core
style cleanup: comment blocks
2012-03-09 18:28:30 +00:00
* - framebuffer object: ext specification
* - multitexture 1.3 core
* - arb non power of two: 2.0 core
* - pixel buffer objects? 2.1 core
* - arb draw buffers? 2.0 core
*/
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
/* Non-generated shaders */
Fix T31546 fragment program gets created every frame That was really crappy indeed. Now we have a separate API for low level OpenGL programs, plus a nice interface for GPU, also removes some GL calls from main code as a plus :) The source for the programs is also moved to nice external .glsl files (not sure which extension convention GPU assemply uses)
2015-04-09 20:20:33 +02:00
extern char datatoc_gpu_program_smoke_frag_glsl[];
extern char datatoc_gpu_program_smoke_color_frag_glsl[];
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
extern char datatoc_gpu_shader_vsm_store_vert_glsl[];
extern char datatoc_gpu_shader_vsm_store_frag_glsl[];
extern char datatoc_gpu_shader_sep_gaussian_blur_vert_glsl[];
extern char datatoc_gpu_shader_sep_gaussian_blur_frag_glsl[];
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
extern char datatoc_gpu_shader_fx_vert_glsl[];
extern char datatoc_gpu_shader_fx_ssao_frag_glsl[];
extern char datatoc_gpu_shader_fx_dof_frag_glsl[];
extern char datatoc_gpu_shader_fx_dof_vert_glsl[];
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
extern char datatoc_gpu_shader_fx_dof_hq_frag_glsl[];
extern char datatoc_gpu_shader_fx_dof_hq_vert_glsl[];
extern char datatoc_gpu_shader_fx_dof_hq_geo_glsl[];
Compositing works with X-Ray. Basically, before drawing X-Rays, we now bind a second depth buffer. After drawing XRays, we do an extra resolve pass where we overwrite the non-XRay depth buffer in pixels where the depth is not maximum (which means background pixel, since depth is cleared before drawing X-Ray objects). This ensures both scene and X-Rays keep their depth values and are ready for compositing. Well, the odd effect due to depth discontinuities can be expected, and X-Rays are a bit more expensive (extra buffer + resolve pass) but at least X-Rays won't invalidate depth values anymore. Whee!
2015-02-17 18:08:28 +01:00
extern char datatoc_gpu_shader_fx_depth_resolve_glsl[];
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
extern char datatoc_gpu_shader_fx_lib_glsl[];
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
typedef struct GPUShaders {
GPUShader *vsm_store;
GPUShader *sep_gaussian_blur;
Fix T31546 fragment program gets created every frame That was really crappy indeed. Now we have a separate API for low level OpenGL programs, plus a nice interface for GPU, also removes some GL calls from main code as a plus :) The source for the programs is also moved to nice external .glsl files (not sure which extension convention GPU assemply uses)
2015-04-09 20:20:33 +02:00
GPUProgram *smoke;
GPUProgram *smoke_colored;
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
/* cache for shader fx. Those can exist in combinations so store them here */
GPUShader *fx_shaders[MAX_FX_SHADERS * 2];
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
} GPUShaders;
Added a -noglsl option to disable GLSL from the command line.
2008-10-12 18:32:26 +00:00
static struct GPUGlobal {
Add function to query maximum texture size. Also, make texture upload functions aware of this limit.
2013-04-12 17:56:07 +00:00
GLint maxtexsize;
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
GLint maxtextures;
GLuint currentfb;
Fix #19669 and other: triple buffer & icon texture drawing could cause a system crash and other issues on ATI/Apple, due to a buggy driver (similar issues reported for other OpenGL applications). For now, work around it by not using non-power-of-two textures on this combination.
2009-10-19 10:10:05 +00:00
int glslsupport;
Added a -noglsl option to disable GLSL from the command line.
2008-10-12 18:32:26 +00:00
int extdisabled;
Fix #21894: backface selection wasn't working correct with < 24 bits colors, e.g. thousands of colors on OS X, due to use of uninitialized value. Problem tracked down and patch provided by Shane Ambler, thanks!
2010-07-04 12:24:19 +00:00
int colordepth;
Detect Gallium driver. Extend NPoT workaround to opensource drivers.
2010-11-27 19:59:00 +00:00
int npotdisabled; /* ATI 3xx-5xx (and more) chipsets support NPoT partially (== not enough) */
BGE: Potential fix for [#35522] Broken game engine compatibility since 2.66a on some ATI cards? Disabling display lists for legacy ATI cards since they don't support display lists well. Also removing an unused variable from the display list rasterizer.
2013-09-11 23:24:45 +00:00
int dlistsdisabled; /* Legacy ATI driver does not support display lists well */
WM Draw Methods now has a new option Automatic (default). This will set the draw method to triple buffer or overlap depending on the configuration. Ideally I could get all cases working well with triple buffer but it's hard in practice. At the moment there are two cases that use overlap instead: * opensource ATI drives on linux * windows software renderer Also added a utility function to check GPU device/os/driver.
2010-01-31 23:45:51 +00:00
GPUDeviceType device;
GPUOSType os;
GPUDriverType driver;
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
GPUShaders shaders;
Fix T40498 invalid textures flickering. Issue here is most likely sampler uniforms and textures not being updated properly when zero binding is created. Solution for now is to allow zero binding but when this happens use sexy pink invalid texture instead :p.
2014-06-10 01:36:53 +03:00
GPUTexture *invalid_tex_1D; /* texture used in place of invalid textures (not loaded correctly, missing) */
GPUTexture *invalid_tex_2D;
GPUTexture *invalid_tex_3D;
Potential fix for T43987, ambient occlusion different between offscreen and on screen rendering. Aaaaah, the beauty of driver implementations of OpenGL! Turns out the problem here is that drivers calculate df/dy differently in some cases (probably because OpenGL counts y reverse to how the window system does, so drivers can get confused). Fixed this for the ATI case based on info we have so far, there's also the Intel case which will be handled separately (missing info on Intel's renderer string etc). Unfortunately we can't really fix this for the general case so we'll have to haldle cases as they come in our tracker and by adding silly string comparisons in our GPU initialization module <sigh>.
2015-03-30 14:14:32 +02:00
float dfdyfactors[2]; /* workaround for different calculation of dfdy factors on GPUs. Some GPUs/drivers
calculate dfdy in shader differently when drawing to an offscreen buffer. First
number is factor on screen and second is off-screen */
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
} GG = {1, 0};
Added a -noglsl option to disable GLSL from the command line.
2008-10-12 18:32:26 +00:00
GPU framebuffer/texture API: Warn when binding a texture that is also attached to a framebuffer or vice versa. might be more correct to just handle the case and unbind here.
2014-11-18 12:12:28 +01:00
/* Number of maximum output slots. We support 4 outputs for now (usually we wouldn't need more to preserve fill rate) */
#define GPU_FB_MAX_SLOTS 4
struct GPUFrameBuffer {
GLuint object;
GPUTexture *colortex[GPU_FB_MAX_SLOTS];
GPUTexture *depthtex;
};
WM Draw Methods now has a new option Automatic (default). This will set the draw method to triple buffer or overlap depending on the configuration. Ideally I could get all cases working well with triple buffer but it's hard in practice. At the moment there are two cases that use overlap instead: * opensource ATI drives on linux * windows software renderer Also added a utility function to check GPU device/os/driver.
2010-01-31 23:45:51 +00:00
/* GPU Types */
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
bool GPU_type_matches(GPUDeviceType device, GPUOSType os, GPUDriverType driver)
WM Draw Methods now has a new option Automatic (default). This will set the draw method to triple buffer or overlap depending on the configuration. Ideally I could get all cases working well with triple buffer but it's hard in practice. At the moment there are two cases that use overlap instead: * opensource ATI drives on linux * windows software renderer Also added a utility function to check GPU device/os/driver.
2010-01-31 23:45:51 +00:00
{
return (GG.device & device) && (GG.os & os) && (GG.driver & driver);
}
/* GPU Extensions */
warning cleanup. - fix mistake with grease pencil UI (&& was intended but & used). - use (void) rather then () across _all_ blenders code. - a few minor edits, don't shadow stack variables in roll calculation & avoid running memset() for VBO vertex map.
2011-02-13 15:02:21 +00:00
void GPU_extensions_disable(void)
Added a -noglsl option to disable GLSL from the command line.
2008-10-12 18:32:26 +00:00
{
GG.extdisabled = 1;
}
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
code cleanup: warnings + style
2013-04-13 00:43:49 +00:00
int GPU_max_texture_size(void)
Add function to query maximum texture size. Also, make texture upload functions aware of this limit.
2013-04-12 17:56:07 +00:00
{
return GG.maxtexsize;
}
Potential fix for T43987, ambient occlusion different between offscreen and on screen rendering. Aaaaah, the beauty of driver implementations of OpenGL! Turns out the problem here is that drivers calculate df/dy differently in some cases (probably because OpenGL counts y reverse to how the window system does, so drivers can get confused). Fixed this for the ATI case based on info we have so far, there's also the Intel case which will be handled separately (missing info on Intel's renderer string etc). Unfortunately we can't really fix this for the general case so we'll have to haldle cases as they come in our tracker and by adding silly string comparisons in our GPU initialization module <sigh>.
2015-03-30 14:14:32 +02:00
void GPU_get_dfdy_factors(float fac[2])
{
copy_v2_v2(fac, GG.dfdyfactors);
}
Ghost Context Refactor https://developer.blender.org/D643 Separates graphics context creation from window code in Ghost so that they can vary separately.
2014-10-07 15:46:19 -05:00
void gpu_extensions_init(void)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
Fix #21894: backface selection wasn't working correct with < 24 bits colors, e.g. thousands of colors on OS X, due to use of uninitialized value. Problem tracked down and patch provided by Shane Ambler, thanks!
2010-07-04 12:24:19 +00:00
GLint r, g, b;
Potential fix for T43987, ambient occlusion different between offscreen and on screen rendering. Aaaaah, the beauty of driver implementations of OpenGL! Turns out the problem here is that drivers calculate df/dy differently in some cases (probably because OpenGL counts y reverse to how the window system does, so drivers can get confused). Fixed this for the ATI case based on info we have so far, there's also the Intel case which will be handled separately (missing info on Intel's renderer string etc). Unfortunately we can't really fix this for the general case so we'll have to haldle cases as they come in our tracker and by adding silly string comparisons in our GPU initialization module <sigh>.
2015-03-30 14:14:32 +02:00
const char *vendor, *renderer, *version;
WM Draw Methods now has a new option Automatic (default). This will set the draw method to triple buffer or overlap depending on the configuration. Ideally I could get all cases working well with triple buffer but it's hard in practice. At the moment there are two cases that use overlap instead: * opensource ATI drives on linux * windows software renderer Also added a utility function to check GPU device/os/driver.
2010-01-31 23:45:51 +00:00
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
/* glewIsSupported("GL_VERSION_2_0") */
if (GLEW_ARB_multitexture)
glGetIntegerv(GL_MAX_TEXTURE_IMAGE_UNITS_ARB, &GG.maxtextures);
Add function to query maximum texture size. Also, make texture upload functions aware of this limit.
2013-04-12 17:56:07 +00:00
glGetIntegerv(GL_MAX_TEXTURE_SIZE, &GG.maxtexsize);
Fix #19669 and other: triple buffer & icon texture drawing could cause a system crash and other issues on ATI/Apple, due to a buggy driver (similar issues reported for other OpenGL applications). For now, work around it by not using non-power-of-two textures on this combination.
2009-10-19 10:10:05 +00:00
GG.glslsupport = 1;
if (!GLEW_ARB_multitexture) GG.glslsupport = 0;
if (!GLEW_ARB_vertex_shader) GG.glslsupport = 0;
if (!GLEW_ARB_fragment_shader) GG.glslsupport = 0;
WM Draw Methods now has a new option Automatic (default). This will set the draw method to triple buffer or overlap depending on the configuration. Ideally I could get all cases working well with triple buffer but it's hard in practice. At the moment there are two cases that use overlap instead: * opensource ATI drives on linux * windows software renderer Also added a utility function to check GPU device/os/driver.
2010-01-31 23:45:51 +00:00
Fix #21894: backface selection wasn't working correct with < 24 bits colors, e.g. thousands of colors on OS X, due to use of uninitialized value. Problem tracked down and patch provided by Shane Ambler, thanks!
2010-07-04 12:24:19 +00:00
glGetIntegerv(GL_RED_BITS, &r);
glGetIntegerv(GL_GREEN_BITS, &g);
glGetIntegerv(GL_BLUE_BITS, &b);
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
GG.colordepth = r + g + b; /* assumes same depth for RGB */
whitespace only, no functional change mixed tabs/spaces --> tabs.
2011-04-21 13:11:51 +00:00
style cleanup
2012-12-28 14:19:05 +00:00
vendor = (const char *)glGetString(GL_VENDOR);
renderer = (const char *)glGetString(GL_RENDERER);
Potential fix for T43987, ambient occlusion different between offscreen and on screen rendering. Aaaaah, the beauty of driver implementations of OpenGL! Turns out the problem here is that drivers calculate df/dy differently in some cases (probably because OpenGL counts y reverse to how the window system does, so drivers can get confused). Fixed this for the ATI case based on info we have so far, there's also the Intel case which will be handled separately (missing info on Intel's renderer string etc). Unfortunately we can't really fix this for the general case so we'll have to haldle cases as they come in our tracker and by adding silly string comparisons in our GPU initialization module <sigh>.
2015-03-30 14:14:32 +02:00
version = (const char *)glGetString(GL_VERSION);
WM Draw Methods now has a new option Automatic (default). This will set the draw method to triple buffer or overlap depending on the configuration. Ideally I could get all cases working well with triple buffer but it's hard in practice. At the moment there are two cases that use overlap instead: * opensource ATI drives on linux * windows software renderer Also added a utility function to check GPU device/os/driver.
2010-01-31 23:45:51 +00:00
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (strstr(vendor, "ATI")) {
WM Draw Methods now has a new option Automatic (default). This will set the draw method to triple buffer or overlap depending on the configuration. Ideally I could get all cases working well with triple buffer but it's hard in practice. At the moment there are two cases that use overlap instead: * opensource ATI drives on linux * windows software renderer Also added a utility function to check GPU device/os/driver.
2010-01-31 23:45:51 +00:00
GG.device = GPU_DEVICE_ATI;
GG.driver = GPU_DRIVER_OFFICIAL;
}
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
else if (strstr(vendor, "NVIDIA")) {
WM Draw Methods now has a new option Automatic (default). This will set the draw method to triple buffer or overlap depending on the configuration. Ideally I could get all cases working well with triple buffer but it's hard in practice. At the moment there are two cases that use overlap instead: * opensource ATI drives on linux * windows software renderer Also added a utility function to check GPU device/os/driver.
2010-01-31 23:45:51 +00:00
GG.device = GPU_DEVICE_NVIDIA;
GG.driver = GPU_DRIVER_OFFICIAL;
}
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
else if (strstr(vendor, "Intel") ||
Make memstat easier to read, add missing Intel ID and cleanups.
2010-04-14 21:12:05 +00:00
/* src/mesa/drivers/dri/intel/intel_context.c */
strstr(renderer, "Mesa DRI Intel") ||
whitespace only, no functional change mixed tabs/spaces --> tabs.
2011-04-21 13:11:51 +00:00
strstr(renderer, "Mesa DRI Mobile Intel")) {
WM Draw Methods now has a new option Automatic (default). This will set the draw method to triple buffer or overlap depending on the configuration. Ideally I could get all cases working well with triple buffer but it's hard in practice. At the moment there are two cases that use overlap instead: * opensource ATI drives on linux * windows software renderer Also added a utility function to check GPU device/os/driver.
2010-01-31 23:45:51 +00:00
GG.device = GPU_DEVICE_INTEL;
GG.driver = GPU_DRIVER_OFFICIAL;
}
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
else if (strstr(renderer, "Mesa DRI R") || (strstr(renderer, "Gallium ") && strstr(renderer, " on ATI "))) {
WM Draw Methods now has a new option Automatic (default). This will set the draw method to triple buffer or overlap depending on the configuration. Ideally I could get all cases working well with triple buffer but it's hard in practice. At the moment there are two cases that use overlap instead: * opensource ATI drives on linux * windows software renderer Also added a utility function to check GPU device/os/driver.
2010-01-31 23:45:51 +00:00
GG.device = GPU_DEVICE_ATI;
GG.driver = GPU_DRIVER_OPENSOURCE;
}
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
else if (strstr(renderer, "Nouveau") || strstr(vendor, "nouveau")) {
WM Draw Methods now has a new option Automatic (default). This will set the draw method to triple buffer or overlap depending on the configuration. Ideally I could get all cases working well with triple buffer but it's hard in practice. At the moment there are two cases that use overlap instead: * opensource ATI drives on linux * windows software renderer Also added a utility function to check GPU device/os/driver.
2010-01-31 23:45:51 +00:00
GG.device = GPU_DEVICE_NVIDIA;
GG.driver = GPU_DRIVER_OPENSOURCE;
}
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
else if (strstr(vendor, "Mesa")) {
WM Draw Methods now has a new option Automatic (default). This will set the draw method to triple buffer or overlap depending on the configuration. Ideally I could get all cases working well with triple buffer but it's hard in practice. At the moment there are two cases that use overlap instead: * opensource ATI drives on linux * windows software renderer Also added a utility function to check GPU device/os/driver.
2010-01-31 23:45:51 +00:00
GG.device = GPU_DEVICE_SOFTWARE;
GG.driver = GPU_DRIVER_SOFTWARE;
}
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
else if (strstr(vendor, "Microsoft")) {
WM Draw Methods now has a new option Automatic (default). This will set the draw method to triple buffer or overlap depending on the configuration. Ideally I could get all cases working well with triple buffer but it's hard in practice. At the moment there are two cases that use overlap instead: * opensource ATI drives on linux * windows software renderer Also added a utility function to check GPU device/os/driver.
2010-01-31 23:45:51 +00:00
GG.device = GPU_DEVICE_SOFTWARE;
GG.driver = GPU_DRIVER_SOFTWARE;
}
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
else if (strstr(renderer, "Apple Software Renderer")) {
WM Draw Methods now has a new option Automatic (default). This will set the draw method to triple buffer or overlap depending on the configuration. Ideally I could get all cases working well with triple buffer but it's hard in practice. At the moment there are two cases that use overlap instead: * opensource ATI drives on linux * windows software renderer Also added a utility function to check GPU device/os/driver.
2010-01-31 23:45:51 +00:00
GG.device = GPU_DEVICE_SOFTWARE;
GG.driver = GPU_DRIVER_SOFTWARE;
}
else {
2.5: automatic draw method now uses overlap for Intel on all platforms.
2010-07-16 10:13:04 +00:00
GG.device = GPU_DEVICE_ANY;
GG.driver = GPU_DRIVER_ANY;
WM Draw Methods now has a new option Automatic (default). This will set the draw method to triple buffer or overlap depending on the configuration. Ideally I could get all cases working well with triple buffer but it's hard in practice. At the moment there are two cases that use overlap instead: * opensource ATI drives on linux * windows software renderer Also added a utility function to check GPU device/os/driver.
2010-01-31 23:45:51 +00:00
}
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (GG.device == GPU_DEVICE_ATI) {
Fix: non-power-of-two-textures were always disabled on ATI/Apple due to poor support in some older cards, now it does a more precise check for problematic cards so that it can be enabled on new cards that do support it properly.
2012-02-04 19:57:09 +00:00
/* ATI 9500 to X2300 cards support NPoT textures poorly
* Incomplete list http://dri.freedesktop.org/wiki/ATIRadeon
* New IDs from MESA's src/gallium/drivers/r300/r300_screen.c
*/
BGE: Potential fix for [#35522] Broken game engine compatibility since 2.66a on some ATI cards? Disabling display lists for legacy ATI cards since they don't support display lists well. Also removing an unused variable from the display list rasterizer.
2013-09-11 23:24:45 +00:00
/* This list is close enough to those using the legacy driver which
* has a bug with display lists and glVertexAttrib
*/
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (strstr(renderer, "R3") || strstr(renderer, "RV3") ||
style cleanup: changes to brace placement / newlines - for/while/if/switch
2012-04-28 06:31:57 +00:00
strstr(renderer, "R4") || strstr(renderer, "RV4") ||
strstr(renderer, "RS4") || strstr(renderer, "RC4") ||
strstr(renderer, "R5") || strstr(renderer, "RV5") ||
strstr(renderer, "RS600") || strstr(renderer, "RS690") ||
strstr(renderer, "RS740") || strstr(renderer, "X1") ||
strstr(renderer, "X2") || strstr(renderer, "Radeon 9") ||
strstr(renderer, "RADEON 9"))
{
Fix: non-power-of-two-textures were always disabled on ATI/Apple due to poor support in some older cards, now it does a more precise check for problematic cards so that it can be enabled on new cards that do support it properly.
2012-02-04 19:57:09 +00:00
GG.npotdisabled = 1;
BGE: Potential fix for [#35522] Broken game engine compatibility since 2.66a on some ATI cards? Disabling display lists for legacy ATI cards since they don't support display lists well. Also removing an unused variable from the display list rasterizer.
2013-09-11 23:24:45 +00:00
GG.dlistsdisabled = 1;
style cleanup: changes to brace placement / newlines - for/while/if/switch
2012-04-28 06:31:57 +00:00
}
Fix: non-power-of-two-textures were always disabled on ATI/Apple due to poor support in some older cards, now it does a more precise check for problematic cards so that it can be enabled on new cards that do support it properly.
2012-02-04 19:57:09 +00:00
}
style cleanup - spelling - turns out we had tessellation spelt wrong all over. - use \directive for doxy (not @directive) - remove BLI_sparsemap.h - was from bmesh merge IIRC but entire file commented and not used.
2012-03-02 16:05:54 +00:00
/* make sure double side isn't used by default and only getting enabled in places where it's
Changes in double side lighting usage Disabled double-side lighting by default and changed areas which are using double sided lighting to switch it off just after doing stuff which requires double side. This makes code a bit more simple to follow (no all that disabling double side lighting all over the code and so). This change also fixes crash of blender caused by intel gme965 driver which leads to stack corruption in some place when double side lighting isn't disabled (probably driver still kind of using double side in some areas or so). Hopefully it'll also fix #30293: Converting Text to Mesh Thanks to Campbell to assist writing this patch!
2012-02-25 18:09:39 +00:00
* really needed to prevent different unexpected behaviors like with intel gme965 card (sergey) */
minor style edits - use GL_TRUE/FALSE for glLightModeli - use set for python checks against multiple string values.
2012-02-25 21:27:45 +00:00
glLightModeli(GL_LIGHT_MODEL_TWO_SIDE, GL_FALSE);
Changes in double side lighting usage Disabled double-side lighting by default and changed areas which are using double sided lighting to switch it off just after doing stuff which requires double side. This makes code a bit more simple to follow (no all that disabling double side lighting all over the code and so). This change also fixes crash of blender caused by intel gme965 driver which leads to stack corruption in some place when double side lighting isn't disabled (probably driver still kind of using double side in some areas or so). Hopefully it'll also fix #30293: Converting Text to Mesh Thanks to Campbell to assist writing this patch!
2012-02-25 18:09:39 +00:00
WM Draw Methods now has a new option Automatic (default). This will set the draw method to triple buffer or overlap depending on the configuration. Ideally I could get all cases working well with triple buffer but it's hard in practice. At the moment there are two cases that use overlap instead: * opensource ATI drives on linux * windows software renderer Also added a utility function to check GPU device/os/driver.
2010-01-31 23:45:51 +00:00
#ifdef _WIN32
GG.os = GPU_OS_WIN;
code cleanup: duplicate checks and double assignments.
2012-03-08 04:38:35 +00:00
#elif defined(__APPLE__)
WM Draw Methods now has a new option Automatic (default). This will set the draw method to triple buffer or overlap depending on the configuration. Ideally I could get all cases working well with triple buffer but it's hard in practice. At the moment there are two cases that use overlap instead: * opensource ATI drives on linux * windows software renderer Also added a utility function to check GPU device/os/driver.
2010-01-31 23:45:51 +00:00
GG.os = GPU_OS_MAC;
code cleanup: duplicate checks and double assignments.
2012-03-08 04:38:35 +00:00
#else
GG.os = GPU_OS_UNIX;
WM Draw Methods now has a new option Automatic (default). This will set the draw method to triple buffer or overlap depending on the configuration. Ideally I could get all cases working well with triple buffer but it's hard in practice. At the moment there are two cases that use overlap instead: * opensource ATI drives on linux * windows software renderer Also added a utility function to check GPU device/os/driver.
2010-01-31 23:45:51 +00:00
#endif
OpenGL: implemenation of fixed function lighting as per pixel GLSL shaders. The code is still unused, but the intention is to use this to solve the double sided lighting problem on NVidia, and to make the materials work on OpenGL ES 2.0 eventually. The code works and matches the fixed function lighting pretty much exactly, but still needs optimizations. The actual integration in object draw will be committed later when more fixing & testing, there's lots of different combinations and unclear OpenGL state here.
2013-02-26 00:49:42 +00:00
Fix T40498 invalid textures flickering. Issue here is most likely sampler uniforms and textures not being updated properly when zero binding is created. Solution for now is to allow zero binding but when this happens use sexy pink invalid texture instead :p.
2014-06-10 01:36:53 +03:00
Potential fix for T43987, ambient occlusion different between offscreen and on screen rendering. Aaaaah, the beauty of driver implementations of OpenGL! Turns out the problem here is that drivers calculate df/dy differently in some cases (probably because OpenGL counts y reverse to how the window system does, so drivers can get confused). Fixed this for the ATI case based on info we have so far, there's also the Intel case which will be handled separately (missing info on Intel's renderer string etc). Unfortunately we can't really fix this for the general case so we'll have to haldle cases as they come in our tracker and by adding silly string comparisons in our GPU initialization module <sigh>.
2015-03-30 14:14:32 +02:00
/* df/dy calculation factors, those are dependent on driver */
if ((strstr(vendor, "ATI") && strstr(version, "3.3.10750"))) {
GG.dfdyfactors[0] = 1.0;
GG.dfdyfactors[1] = -1.0;
}
Final solution for Intel card Ambient Occlusion in T43987. Forgot to account for offscreen case in addition to compositing
2015-05-18 22:05:38 +02:00
else if (GG.device == GPU_DEVICE_INTEL && GG.os == GPU_OS_WIN &&
Yet another crappy Intel driver added for correct SSAO support.
2015-05-21 18:37:48 +02:00
(strstr(version, "4.0.0 - Build 10.18.10.3308") ||
strstr(version, "4.0.0 - Build 9.18.10.3186") ||
Fix part of T44768 Yet another Intel driver failing on viewport Ambient occlusion (Maybe we should add drivers that DO work instead, but in the future those will be much more since Intel claims to have fixed the issue)
2015-05-19 16:25:42 +02:00
strstr(version, "4.0.0 - Build 9.18.10.3165") ||
Final solution for Intel card Ambient Occlusion in T43987. Forgot to account for offscreen case in addition to compositing
2015-05-18 22:05:38 +02:00
strstr(version, "3.1.0 - Build 9.17.10.3347") ||
strstr(version, "3.1.0 - Build 9.17.10.4101")))
{
GG.dfdyfactors[0] = -1.0;
GG.dfdyfactors[1] = 1.0;
Potential fix for T43987, ambient occlusion different between offscreen and on screen rendering. Aaaaah, the beauty of driver implementations of OpenGL! Turns out the problem here is that drivers calculate df/dy differently in some cases (probably because OpenGL counts y reverse to how the window system does, so drivers can get confused). Fixed this for the ATI case based on info we have so far, there's also the Intel case which will be handled separately (missing info on Intel's renderer string etc). Unfortunately we can't really fix this for the general case so we'll have to haldle cases as they come in our tracker and by adding silly string comparisons in our GPU initialization module <sigh>.
2015-03-30 14:14:32 +02:00
}
else {
GG.dfdyfactors[0] = 1.0;
GG.dfdyfactors[1] = 1.0;
}
Fix T40498 invalid textures flickering. Issue here is most likely sampler uniforms and textures not being updated properly when zero binding is created. Solution for now is to allow zero binding but when this happens use sexy pink invalid texture instead :p.
2014-06-10 01:36:53 +03:00
GPU_invalid_tex_init();
OpenGL: more work on fixed function lighting implementation as GLSL. * Rename functions and move to own header. * Add wrapper functions for glLight. * Auto detect if we can use faster code for solid lighting. * Various fixes for textured draw mode.
2013-03-10 15:38:23 +00:00
GPU_simple_shaders_init();
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
Ghost Context Refactor https://developer.blender.org/D643 Separates graphics context creation from window code in Ghost so that they can vary separately.
2014-10-07 15:46:19 -05:00
void gpu_extensions_exit(void)
Cleaning up the GPU_extensions_init/exit() code a bit to keep the Blenderplayer from crashing on exit and restart.
2011-12-23 07:10:01 +00:00
{
OpenGL: more work on fixed function lighting implementation as GLSL. * Rename functions and move to own header. * Add wrapper functions for glLight. * Auto detect if we can use faster code for solid lighting. * Various fixes for textured draw mode.
2013-03-10 15:38:23 +00:00
GPU_simple_shaders_exit();
Fix T40498 invalid textures flickering. Issue here is most likely sampler uniforms and textures not being updated properly when zero binding is created. Solution for now is to allow zero binding but when this happens use sexy pink invalid texture instead :p.
2014-06-10 01:36:53 +03:00
GPU_invalid_tex_free();
Cleaning up the GPU_extensions_init/exit() code a bit to keep the Blenderplayer from crashing on exit and restart.
2011-12-23 07:10:01 +00:00
}
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
bool GPU_glsl_support(void)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
Fix #19669 and other: triple buffer & icon texture drawing could cause a system crash and other issues on ATI/Apple, due to a buggy driver (similar issues reported for other OpenGL applications). For now, work around it by not using non-power-of-two textures on this combination.
2009-10-19 10:10:05 +00:00
return !GG.extdisabled && GG.glslsupport;
}
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
bool GPU_non_power_of_two_support(void)
Fix #19669 and other: triple buffer & icon texture drawing could cause a system crash and other issues on ATI/Apple, due to a buggy driver (similar issues reported for other OpenGL applications). For now, work around it by not using non-power-of-two textures on this combination.
2009-10-19 10:10:05 +00:00
{
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (GG.npotdisabled)
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
return false;
ATI X1xxx gfx cards (R500 chipset) lack full support for npot textures although they report the GLEW_ARB_texture_non_power_of_two extension.
2010-11-25 22:15:04 +00:00
Fix #19669 and other: triple buffer & icon texture drawing could cause a system crash and other issues on ATI/Apple, due to a buggy driver (similar issues reported for other OpenGL applications). For now, work around it by not using non-power-of-two textures on this combination.
2009-10-19 10:10:05 +00:00
return GLEW_ARB_texture_non_power_of_two;
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
bool GPU_vertex_buffer_support(void)
Another stability commit for legacy users - don't crash when toggling dyntopo just notify user that their system doesn't support it.
2015-03-03 11:34:23 +01:00
{
return GLEW_ARB_vertex_buffer_object || GLEW_VERSION_1_5;
}
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
bool GPU_display_list_support(void)
BGE: Potential fix for [#35522] Broken game engine compatibility since 2.66a on some ATI cards? Disabling display lists for legacy ATI cards since they don't support display lists well. Also removing an unused variable from the display list rasterizer.
2013-09-11 23:24:45 +00:00
{
return !GG.dlistsdisabled;
}
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
bool GPU_bicubic_bump_support(void)
{
return GLEW_ARB_texture_query_lod && GLEW_VERSION_3_0;
}
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
bool GPU_geometry_shader_support(void)
{
return GLEW_EXT_geometry_shader4 || GLEW_VERSION_3_2;
}
bool GPU_instanced_drawing_support(void)
{
Fix high quality depth of field on the Mac. Quite a few things wrong here: * Mac did not support EXT_draw_instanced, only ARB_draw_instanced * Draw instanced did not work unless data came from vertex buffer, which is second time we see weird things with vertex arrays in mac * There were a few stupid mistakes by me as well, such as binding to uniform locations for the wrong shaders (it's a wonder it ever worked :p)
2015-03-20 15:26:13 +01:00
return GLEW_ARB_draw_instanced;
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
}
warning cleanup. - fix mistake with grease pencil UI (&& was intended but & used). - use (void) rather then () across _all_ blenders code. - a few minor edits, don't shadow stack variables in roll calculation & avoid running memset() for VBO vertex map.
2011-02-13 15:02:21 +00:00
int GPU_color_depth(void)
Fix #21349: triple buffer drawing doesn't work well with thousands of colors setting on Mac, just disabled it in that case.
2010-04-05 10:25:40 +00:00
{
whitespace only, no functional change mixed tabs/spaces --> tabs.
2011-04-21 13:11:51 +00:00
return GG.colordepth;
Fix #21349: triple buffer drawing doesn't work well with thousands of colors setting on Mac, just disabled it in that case.
2010-04-05 10:25:40 +00:00
}
bug [#26329] Project Paint not working we cant ensure that a requested buffer can be allocated so report opengl errors when failing to allocate the buffer (rather then printing to console). this is common enough and generic error isn't too helpful to users.
2011-03-06 23:12:12 +00:00
static void GPU_print_framebuffer_error(GLenum status, char err_out[256])
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
const char *err = "unknown";
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
style cleanup: changes to brace placement / newlines - for/while/if/switch
2012-04-28 06:31:57 +00:00
switch (status) {
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
case GL_FRAMEBUFFER_COMPLETE_EXT:
break;
Fix #33371: blender freezing in material draw mode. When FBO failed in a particular way it could cause the opengl draw buffer to be set wrong, effectively disabling all opengl drawing. The FBO error was caused by cycles GLSL materials with no nodes that would still use blender internal materials, which caused issues with lamp shadow buffers FBO. This also fixes a GLSL refresh issue when switching render engines.
2012-12-03 08:31:16 +00:00
case GL_INVALID_OPERATION:
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
err = "Invalid operation";
Fix #33371: blender freezing in material draw mode. When FBO failed in a particular way it could cause the opengl draw buffer to be set wrong, effectively disabling all opengl drawing. The FBO error was caused by cycles GLSL materials with no nodes that would still use blender internal materials, which caused issues with lamp shadow buffers FBO. This also fixes a GLSL refresh issue when switching render engines.
2012-12-03 08:31:16 +00:00
break;
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
case GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT_EXT:
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
err = "Incomplete attachment";
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
break;
case GL_FRAMEBUFFER_UNSUPPORTED_EXT:
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
err = "Unsupported framebuffer format";
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
break;
case GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT_EXT:
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
err = "Missing attachment";
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
break;
case GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT:
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
err = "Attached images must have same dimensions";
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
break;
case GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT:
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
err = "Attached images must have same format";
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
break;
case GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT:
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
err = "Missing draw buffer";
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
break;
case GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT:
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
err = "Missing read buffer";
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
break;
}
bug [#26329] Project Paint not working we cant ensure that a requested buffer can be allocated so report opengl errors when failing to allocate the buffer (rather then printing to console). this is common enough and generic error isn't too helpful to users.
2011-03-06 23:12:12 +00:00
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (err_out) {
se BLI_snprintf() for msvc compat, cleanup CMake file, some bad comments left in. also on only try build RPM's on linux.
2011-03-07 03:57:04 +00:00
BLI_snprintf(err_out, 256, "GPUFrameBuffer: framebuffer incomplete error %d '%s'",
(int)status, err);
bug [#26329] Project Paint not working we cant ensure that a requested buffer can be allocated so report opengl errors when failing to allocate the buffer (rather then printing to console). this is common enough and generic error isn't too helpful to users.
2011-03-06 23:12:12 +00:00
}
else {
fprintf(stderr, "GPUFrameBuffer: framebuffer incomplete error %d '%s'\n",
se BLI_snprintf() for msvc compat, cleanup CMake file, some bad comments left in. also on only try build RPM's on linux.
2011-03-07 03:57:04 +00:00
(int)status, err);
bug [#26329] Project Paint not working we cant ensure that a requested buffer can be allocated so report opengl errors when failing to allocate the buffer (rather then printing to console). this is common enough and generic error isn't too helpful to users.
2011-03-06 23:12:12 +00:00
}
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
/* GPUTexture */
struct GPUTexture {
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
int w, h; /* width/height */
Alternate fix for offscreen render w/o npot Store the original texture size, since its needed for reading pixels and passing to glViewport.
2015-09-14 13:18:43 +10:00
int w_orig, h_orig; /* width/height (before power of 2 is applied) */
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
int number; /* number for multitexture binding */
int refcount; /* reference count */
GLenum target; /* GL_TEXTURE_* */
GLuint bindcode; /* opengl identifier for texture */
int fromblender; /* we got the texture from Blender */
GPUFrameBuffer *fb; /* GPUFramebuffer this texture is attached to */
int fb_attachment; /* slot the texture is attached to */
int depth; /* is a depth texture? if 3D how deep? */
Alternate fix for offscreen render w/o npot Store the original texture size, since its needed for reading pixels and passing to glViewport.
2015-09-14 13:18:43 +10:00
int depth_orig; /* depth (before power of 2 is applied) */
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
};
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
static unsigned char *GPU_texture_convert_pixels(int length, const float *fpixels)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
unsigned char *pixels, *p;
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
const float *fp = fpixels;
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
const int len = 4 * length;
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
int a;
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
p = pixels = MEM_callocN(sizeof(unsigned char) * len, "GPUTexturePixels");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
for (a = 0; a < len; a++, p++, fp++)
*p = FTOCHAR((*fp));
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
return pixels;
}
static void GPU_glTexSubImageEmpty(GLenum target, GLenum format, int x, int y, int w, int h)
{
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
void *pixels = MEM_callocN(sizeof(char) * 4 * w * h, "GPUTextureEmptyPixels");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
if (target == GL_TEXTURE_1D)
glTexSubImage1D(target, 0, x, w, format, GL_UNSIGNED_BYTE, pixels);
else
glTexSubImage2D(target, 0, x, y, w, h, format, GL_UNSIGNED_BYTE, pixels);
MEM_freeN(pixels);
}
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
static GPUTexture *GPU_texture_create_nD(
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
int w, int h, int n, const float *fpixels, int depth, GPUHDRType hdr_type, int components,
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
char err_out[256])
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
GPUTexture *tex;
GLenum type, format, internalformat;
void *pixels = NULL;
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (depth && !GLEW_ARB_depth_texture)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
return NULL;
tex = MEM_callocN(sizeof(GPUTexture), "GPUTexture");
Alternate fix for offscreen render w/o npot Store the original texture size, since its needed for reading pixels and passing to glViewport.
2015-09-14 13:18:43 +10:00
tex->w = tex->w_orig = w;
tex->h = tex->h_orig = h;
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
tex->number = -1;
tex->refcount = 1;
tex->target = (n == 1)? GL_TEXTURE_1D: GL_TEXTURE_2D;
Alternate fix for offscreen render w/o npot Store the original texture size, since its needed for reading pixels and passing to glViewport.
2015-09-14 13:18:43 +10:00
tex->depth = tex->depth_orig = depth;
GPU framebuffer API: Allow binding a texture to a different texture attachment than the first. Also fix a number error in seperable gaussian blur shader.
2014-11-18 11:56:23 +01:00
tex->fb_attachment = -1;
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
glGenTextures(1, &tex->bindcode);
if (!tex->bindcode) {
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (err_out) {
se BLI_snprintf() for msvc compat, cleanup CMake file, some bad comments left in. also on only try build RPM's on linux.
2011-03-07 03:57:04 +00:00
BLI_snprintf(err_out, 256, "GPUTexture: texture create failed: %d",
bug [#26329] Project Paint not working we cant ensure that a requested buffer can be allocated so report opengl errors when failing to allocate the buffer (rather then printing to console). this is common enough and generic error isn't too helpful to users.
2011-03-06 23:12:12 +00:00
(int)glGetError());
}
else {
fprintf(stderr, "GPUTexture: texture create failed: %d\n",
(int)glGetError());
}
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
GPU_texture_free(tex);
return NULL;
}
Fix #19669 and other: triple buffer & icon texture drawing could cause a system crash and other issues on ATI/Apple, due to a buggy driver (similar issues reported for other OpenGL applications). For now, work around it by not using non-power-of-two textures on this combination.
2009-10-19 10:10:05 +00:00
if (!GPU_non_power_of_two_support()) {
static functions for getting power of 2 values were being copied about too much, add to the BLI_math api. - is_power_of_2_i - power_of_2_min_i - power_of_2_max_i
2011-12-16 09:25:07 +00:00
tex->w = power_of_2_max_i(tex->w);
tex->h = power_of_2_max_i(tex->h);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
tex->number = 0;
glBindTexture(tex->target, tex->bindcode);
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (depth) {
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
type = GL_UNSIGNED_BYTE;
format = GL_DEPTH_COMPONENT;
internalformat = GL_DEPTH_COMPONENT;
}
else {
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
type = GL_FLOAT;
if (components == 4) {
format = GL_RGBA;
switch (hdr_type) {
case GPU_HDR_NONE:
internalformat = GL_RGBA8;
break;
case GPU_HDR_HALF_FLOAT:
internalformat = GL_RGBA16F;
break;
case GPU_HDR_FULL_FLOAT:
internalformat = GL_RGBA32F;
break;
default:
break;
}
}
else if (components == 2) {
format = GL_RG;
switch (hdr_type) {
case GPU_HDR_NONE:
internalformat = GL_RG8;
break;
case GPU_HDR_HALF_FLOAT:
internalformat = GL_RG16F;
break;
case GPU_HDR_FULL_FLOAT:
internalformat = GL_RG32F;
break;
default:
break;
}
}
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
if (fpixels && hdr_type == GPU_HDR_NONE) {
type = GL_UNSIGNED_BYTE;
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
pixels = GPU_texture_convert_pixels(w*h, fpixels);
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
}
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
if (tex->target == GL_TEXTURE_1D) {
use NULL rather then 0 for pointer assignments & comparison, modifier, imbuf & editors.
2011-03-05 10:29:10 +00:00
glTexImage1D(tex->target, 0, internalformat, tex->w, 0, format, type, NULL);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
if (fpixels) {
glTexSubImage1D(tex->target, 0, 0, w, format, type,
style cleanup
2013-03-31 03:28:46 +00:00
pixels ? pixels : fpixels);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
if (tex->w > w)
GPU_glTexSubImageEmpty(tex->target, format, w, 0,
tex->w-w, 1);
}
}
else {
glTexImage2D(tex->target, 0, internalformat, tex->w, tex->h, 0,
style cleanup: osl and NULL pointer use, also correct sequencer gap operator id's
2013-03-27 20:27:07 +00:00
format, type, NULL);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
if (fpixels) {
glTexSubImage2D(tex->target, 0, 0, 0, w, h,
style cleanup
2013-03-31 03:28:46 +00:00
format, type, pixels ? pixels : fpixels);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
if (tex->w > w)
GPU_glTexSubImageEmpty(tex->target, format, w, 0, tex->w-w, tex->h);
if (tex->h > h)
GPU_glTexSubImageEmpty(tex->target, format, 0, h, w, tex->h-h);
}
}
if (pixels)
MEM_freeN(pixels);
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (depth) {
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
glTexParameteri(tex->target, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
glTexParameteri(tex->target, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
glTexParameteri(tex->target, GL_TEXTURE_COMPARE_MODE_ARB, GL_COMPARE_R_TO_TEXTURE);
glTexParameteri(tex->target, GL_TEXTURE_COMPARE_FUNC_ARB, GL_LEQUAL);
glTexParameteri(tex->target, GL_DEPTH_TEXTURE_MODE_ARB, GL_INTENSITY);
}
else {
glTexParameteri(tex->target, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
glTexParameteri(tex->target, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
}
if (tex->target != GL_TEXTURE_1D) {
Remove border clamping. Usually we just want to clamp to edge instead. Note - checked all glTexImage functions and we never use that. Border is ifdefed out too here.
2015-01-05 14:23:15 +01:00
glTexParameteri(tex->target, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
glTexParameteri(tex->target, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
else
glTexParameteri(tex->target, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
return tex;
}
Smoke: * cache for low res (deactivating high res for now) * new way of view3d rendering of smoke (no longer 3 axes) -using 3dtexture now (introduced into gpu/intern) * introducing LZO and LZMA libs into extern (makefiles missing for now) * reducing memory usage after simulating for the frame ended (freeing temporary buffers) * splitting smoke into 2 modifier for the cache-sake (it cannot handle more than 1 cache on the same modifier-index) * no color on gui anymore * fixing non-power-of-2 resolutions (hopefully) * fixing select-deselect of domain drawing bug * fixing drawobject.c coding style (making Ton happy) ;-) HINT #1: If scons doesn't work -> cmakefiles are up-to-date, couldn't test scons (but i tried to mantain them, too) CODERS HINT #1: we really need a way to disable adding all modifiers through "Add Modifiers" dropdown! WARNING #1: before applying this commit, deactivate your SMOKE DOMAIN in your old files and save them then. You can open them then savely after that. WARNING #2: File and cache format of smoke can be changed, this is not final!
2009-08-20 00:33:59 +00:00
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
GPUTexture *GPU_texture_create_3D(int w, int h, int depth, int channels, const float *fpixels)
Smoke: * cache for low res (deactivating high res for now) * new way of view3d rendering of smoke (no longer 3 axes) -using 3dtexture now (introduced into gpu/intern) * introducing LZO and LZMA libs into extern (makefiles missing for now) * reducing memory usage after simulating for the frame ended (freeing temporary buffers) * splitting smoke into 2 modifier for the cache-sake (it cannot handle more than 1 cache on the same modifier-index) * no color on gui anymore * fixing non-power-of-2 resolutions (hopefully) * fixing select-deselect of domain drawing bug * fixing drawobject.c coding style (making Ton happy) ;-) HINT #1: If scons doesn't work -> cmakefiles are up-to-date, couldn't test scons (but i tried to mantain them, too) CODERS HINT #1: we really need a way to disable adding all modifiers through "Add Modifiers" dropdown! WARNING #1: before applying this commit, deactivate your SMOKE DOMAIN in your old files and save them then. You can open them then savely after that. WARNING #2: File and cache format of smoke can be changed, this is not final!
2009-08-20 00:33:59 +00:00
{
GPUTexture *tex;
GLenum type, format, internalformat;
void *pixels = NULL;
3D textures: detect if we are pushing OpenGL too hard with the texture size.
2015-04-23 19:02:08 +02:00
int r_width;
3D Textures: If we fail allocating a proxy texture don't fail, instead create a smaller nearest filtered image to display in its place. This can make viewing slow (it's an extra O^3 operation), but this will probably help us render the tornado in 3D viewport in gooseberry and still actually see something - despite the rendering taking longer. I've added a debug print so we can know when this happens.
2015-04-24 11:35:49 +02:00
bool rescale = false;
Smoke: * cache for low res (deactivating high res for now) * new way of view3d rendering of smoke (no longer 3 axes) -using 3dtexture now (introduced into gpu/intern) * introducing LZO and LZMA libs into extern (makefiles missing for now) * reducing memory usage after simulating for the frame ended (freeing temporary buffers) * splitting smoke into 2 modifier for the cache-sake (it cannot handle more than 1 cache on the same modifier-index) * no color on gui anymore * fixing non-power-of-2 resolutions (hopefully) * fixing select-deselect of domain drawing bug * fixing drawobject.c coding style (making Ton happy) ;-) HINT #1: If scons doesn't work -> cmakefiles are up-to-date, couldn't test scons (but i tried to mantain them, too) CODERS HINT #1: we really need a way to disable adding all modifiers through "Add Modifiers" dropdown! WARNING #1: before applying this commit, deactivate your SMOKE DOMAIN in your old files and save them then. You can open them then savely after that. WARNING #2: File and cache format of smoke can be changed, this is not final!
2009-08-20 00:33:59 +00:00
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (!GLEW_VERSION_1_2)
Fix #21667: smoke drawing crashes calling glTexImage3D on graphics card that do not support it.
2010-03-27 10:48:01 +00:00
return NULL;
Smoke: * cache for low res (deactivating high res for now) * new way of view3d rendering of smoke (no longer 3 axes) -using 3dtexture now (introduced into gpu/intern) * introducing LZO and LZMA libs into extern (makefiles missing for now) * reducing memory usage after simulating for the frame ended (freeing temporary buffers) * splitting smoke into 2 modifier for the cache-sake (it cannot handle more than 1 cache on the same modifier-index) * no color on gui anymore * fixing non-power-of-2 resolutions (hopefully) * fixing select-deselect of domain drawing bug * fixing drawobject.c coding style (making Ton happy) ;-) HINT #1: If scons doesn't work -> cmakefiles are up-to-date, couldn't test scons (but i tried to mantain them, too) CODERS HINT #1: we really need a way to disable adding all modifiers through "Add Modifiers" dropdown! WARNING #1: before applying this commit, deactivate your SMOKE DOMAIN in your old files and save them then. You can open them then savely after that. WARNING #2: File and cache format of smoke can be changed, this is not final!
2009-08-20 00:33:59 +00:00
tex = MEM_callocN(sizeof(GPUTexture), "GPUTexture");
Alternate fix for offscreen render w/o npot Store the original texture size, since its needed for reading pixels and passing to glViewport.
2015-09-14 13:18:43 +10:00
tex->w = tex->w_orig = w;
tex->h = tex->h_orig = h;
tex->depth = tex->depth_orig = depth;
Smoke: * cache for low res (deactivating high res for now) * new way of view3d rendering of smoke (no longer 3 axes) -using 3dtexture now (introduced into gpu/intern) * introducing LZO and LZMA libs into extern (makefiles missing for now) * reducing memory usage after simulating for the frame ended (freeing temporary buffers) * splitting smoke into 2 modifier for the cache-sake (it cannot handle more than 1 cache on the same modifier-index) * no color on gui anymore * fixing non-power-of-2 resolutions (hopefully) * fixing select-deselect of domain drawing bug * fixing drawobject.c coding style (making Ton happy) ;-) HINT #1: If scons doesn't work -> cmakefiles are up-to-date, couldn't test scons (but i tried to mantain them, too) CODERS HINT #1: we really need a way to disable adding all modifiers through "Add Modifiers" dropdown! WARNING #1: before applying this commit, deactivate your SMOKE DOMAIN in your old files and save them then. You can open them then savely after that. WARNING #2: File and cache format of smoke can be changed, this is not final!
2009-08-20 00:33:59 +00:00
tex->number = -1;
tex->refcount = 1;
tex->target = GL_TEXTURE_3D;
glGenTextures(1, &tex->bindcode);
if (!tex->bindcode) {
fprintf(stderr, "GPUTexture: texture create failed: %d\n",
(int)glGetError());
GPU_texture_free(tex);
return NULL;
}
Fix #19669 and other: triple buffer & icon texture drawing could cause a system crash and other issues on ATI/Apple, due to a buggy driver (similar issues reported for other OpenGL applications). For now, work around it by not using non-power-of-two textures on this combination.
2009-10-19 10:10:05 +00:00
if (!GPU_non_power_of_two_support()) {
static functions for getting power of 2 values were being copied about too much, add to the BLI_math api. - is_power_of_2_i - power_of_2_min_i - power_of_2_max_i
2011-12-16 09:25:07 +00:00
tex->w = power_of_2_max_i(tex->w);
tex->h = power_of_2_max_i(tex->h);
tex->depth = power_of_2_max_i(tex->depth);
Smoke: * cache for low res (deactivating high res for now) * new way of view3d rendering of smoke (no longer 3 axes) -using 3dtexture now (introduced into gpu/intern) * introducing LZO and LZMA libs into extern (makefiles missing for now) * reducing memory usage after simulating for the frame ended (freeing temporary buffers) * splitting smoke into 2 modifier for the cache-sake (it cannot handle more than 1 cache on the same modifier-index) * no color on gui anymore * fixing non-power-of-2 resolutions (hopefully) * fixing select-deselect of domain drawing bug * fixing drawobject.c coding style (making Ton happy) ;-) HINT #1: If scons doesn't work -> cmakefiles are up-to-date, couldn't test scons (but i tried to mantain them, too) CODERS HINT #1: we really need a way to disable adding all modifiers through "Add Modifiers" dropdown! WARNING #1: before applying this commit, deactivate your SMOKE DOMAIN in your old files and save them then. You can open them then savely after that. WARNING #2: File and cache format of smoke can be changed, this is not final!
2009-08-20 00:33:59 +00:00
}
tex->number = 0;
glBindTexture(tex->target, tex->bindcode);
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_ASSERT_NO_GL_ERRORS("3D glBindTexture");
Smoke: * Less verbose * More OpenGL error messages (on blender -d)
2009-09-15 10:26:36 +00:00
Google Summer of Code project: "Smoke Simulator Improvements & Fire". Documentation & Test blend files: ------------------ http://wiki.blender.org/index.php/User:MiikaH/GSoC-2012-Smoke-Simulator-Improvements Credits: ------------------ Miika Hamalainen (MiikaH): Student / Main programmer Daniel Genrich (Genscher): Mentor / Programmer of merged patches from Smoke2 branch Google: For Google Summer of Code 2012
2012-10-10 13:18:07 +00:00
type = GL_FLOAT;
if (channels == 4) {
format = GL_RGBA;
internalformat = GL_RGBA;
}
else {
format = GL_RED;
internalformat = GL_INTENSITY;
}
Smoke: * cache for low res (deactivating high res for now) * new way of view3d rendering of smoke (no longer 3 axes) -using 3dtexture now (introduced into gpu/intern) * introducing LZO and LZMA libs into extern (makefiles missing for now) * reducing memory usage after simulating for the frame ended (freeing temporary buffers) * splitting smoke into 2 modifier for the cache-sake (it cannot handle more than 1 cache on the same modifier-index) * no color on gui anymore * fixing non-power-of-2 resolutions (hopefully) * fixing select-deselect of domain drawing bug * fixing drawobject.c coding style (making Ton happy) ;-) HINT #1: If scons doesn't work -> cmakefiles are up-to-date, couldn't test scons (but i tried to mantain them, too) CODERS HINT #1: we really need a way to disable adding all modifiers through "Add Modifiers" dropdown! WARNING #1: before applying this commit, deactivate your SMOKE DOMAIN in your old files and save them then. You can open them then savely after that. WARNING #2: File and cache format of smoke can be changed, this is not final!
2009-08-20 00:33:59 +00:00
3D textures: detect if we are pushing OpenGL too hard with the texture size.
2015-04-23 19:02:08 +02:00
/* 3D textures are quite heavy, test if it's possible to create them first */
glTexImage3D(GL_PROXY_TEXTURE_3D, 0, internalformat, tex->w, tex->h, tex->depth, 0, format, type, NULL);
glGetTexLevelParameteriv(GL_PROXY_TEXTURE_3D, 0, GL_TEXTURE_WIDTH, &r_width);
3D Textures: If we fail allocating a proxy texture don't fail, instead create a smaller nearest filtered image to display in its place. This can make viewing slow (it's an extra O^3 operation), but this will probably help us render the tornado in 3D viewport in gooseberry and still actually see something - despite the rendering taking longer. I've added a debug print so we can know when this happens.
2015-04-24 11:35:49 +02:00
while (r_width == 0) {
rescale = true;
tex->w /= 2;
tex->h /= 2;
tex->depth /= 2;
glTexImage3D(GL_PROXY_TEXTURE_3D, 0, internalformat, tex->w, tex->h, tex->depth, 0, format, type, NULL);
glGetTexLevelParameteriv(GL_PROXY_TEXTURE_3D, 0, GL_TEXTURE_WIDTH, &r_width);
3D textures: detect if we are pushing OpenGL too hard with the texture size.
2015-04-23 19:02:08 +02:00
}
3D Textures: If we fail allocating a proxy texture don't fail, instead create a smaller nearest filtered image to display in its place. This can make viewing slow (it's an extra O^3 operation), but this will probably help us render the tornado in 3D viewport in gooseberry and still actually see something - despite the rendering taking longer. I've added a debug print so we can know when this happens.
2015-04-24 11:35:49 +02:00
/* really unlikely to happen but keep this just in case */
tex->w = max_ii(tex->w, 1);
tex->h = max_ii(tex->h, 1);
tex->depth = max_ii(tex->depth, 1);
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
#if 0
if (fpixels)
pixels = GPU_texture_convert_pixels(w*h*depth, fpixels);
#endif
Smoke: * cache for low res (deactivating high res for now) * new way of view3d rendering of smoke (no longer 3 axes) -using 3dtexture now (introduced into gpu/intern) * introducing LZO and LZMA libs into extern (makefiles missing for now) * reducing memory usage after simulating for the frame ended (freeing temporary buffers) * splitting smoke into 2 modifier for the cache-sake (it cannot handle more than 1 cache on the same modifier-index) * no color on gui anymore * fixing non-power-of-2 resolutions (hopefully) * fixing select-deselect of domain drawing bug * fixing drawobject.c coding style (making Ton happy) ;-) HINT #1: If scons doesn't work -> cmakefiles are up-to-date, couldn't test scons (but i tried to mantain them, too) CODERS HINT #1: we really need a way to disable adding all modifiers through "Add Modifiers" dropdown! WARNING #1: before applying this commit, deactivate your SMOKE DOMAIN in your old files and save them then. You can open them then savely after that. WARNING #2: File and cache format of smoke can be changed, this is not final!
2009-08-20 00:33:59 +00:00
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_ASSERT_NO_GL_ERRORS("3D glTexImage3D");
Smoke: * Less verbose * More OpenGL error messages (on blender -d)
2009-09-15 10:26:36 +00:00
3D Textures: If we fail allocating a proxy texture don't fail, instead create a smaller nearest filtered image to display in its place. This can make viewing slow (it's an extra O^3 operation), but this will probably help us render the tornado in 3D viewport in gooseberry and still actually see something - despite the rendering taking longer. I've added a debug print so we can know when this happens.
2015-04-24 11:35:49 +02:00
/* hardcore stuff, 3D texture rescaling - warning, this is gonna hurt your performance a lot, but we need it
* for gooseberry */
if (rescale && fpixels) {
unsigned int i, j, k;
unsigned int xf = w / tex->w, yf = h / tex->h, zf = depth / tex->depth;
float *tex3d = MEM_mallocN(channels * sizeof(float)*tex->w*tex->h*tex->depth, "tex3d");
GPU_print_error_debug("You need to scale a 3D texture, feel the pain!");
for (k = 0; k < tex->depth; k++) {
for (j = 0; j < tex->h; j++) {
for (i = 0; i < tex->w; i++) {
/* obviously doing nearest filtering here, it's going to be slow in any case, let's not make it worse */
float xb = i * xf;
float yb = j * yf;
float zb = k * zf;
unsigned int offset = k * (tex->w * tex->h) + i * tex->h + j;
unsigned int offset_orig = (zb) * (w * h) + (xb) * h + (yb);
if (channels == 4) {
tex3d[offset * 4] = fpixels[offset_orig * 4];
tex3d[offset * 4 + 1] = fpixels[offset_orig * 4 + 1];
tex3d[offset * 4 + 2] = fpixels[offset_orig * 4 + 2];
tex3d[offset * 4 + 3] = fpixels[offset_orig * 4 + 3];
}
else
tex3d[offset] = fpixels[offset_orig];
}
}
3D textures: detect if we are pushing OpenGL too hard with the texture size.
2015-04-23 19:02:08 +02:00
}
Fix #19875: drawing smoke on graphics cards that do not support non-power-of-two textures lead to artifacts due to uninitialized memory if the domain had a non-power-of-two size.
2010-01-30 09:24:50 +00:00
3D Textures: If we fail allocating a proxy texture don't fail, instead create a smaller nearest filtered image to display in its place. This can make viewing slow (it's an extra O^3 operation), but this will probably help us render the tornado in 3D viewport in gooseberry and still actually see something - despite the rendering taking longer. I've added a debug print so we can know when this happens.
2015-04-24 11:35:49 +02:00
glTexImage3D(tex->target, 0, internalformat, tex->w, tex->h, tex->depth, 0, format, type, tex3d);
MEM_freeN(tex3d);
}
else {
if (fpixels) {
if (!GPU_non_power_of_two_support() && (w != tex->w || h != tex->h || depth != tex->depth)) {
/* clear first to avoid unitialized pixels */
float *zero= MEM_callocN(sizeof(float)*tex->w*tex->h*tex->depth, "zero");
glTexImage3D(tex->target, 0, internalformat, tex->w, tex->h, tex->depth, 0, format, type, NULL);
glTexSubImage3D(tex->target, 0, 0, 0, 0, tex->w, tex->h, tex->depth, GL_INTENSITY, GL_FLOAT, zero);
glTexSubImage3D(tex->target, 0, 0, 0, 0, w, h, depth, format, type, fpixels);
MEM_freeN(zero);
}
else {
glTexImage3D(tex->target, 0, internalformat, tex->w, tex->h, tex->depth, 0, format, type, fpixels);
}
GPU_ASSERT_NO_GL_ERRORS("3D glTexSubImage3D");
}
Smoke: * cache for low res (deactivating high res for now) * new way of view3d rendering of smoke (no longer 3 axes) -using 3dtexture now (introduced into gpu/intern) * introducing LZO and LZMA libs into extern (makefiles missing for now) * reducing memory usage after simulating for the frame ended (freeing temporary buffers) * splitting smoke into 2 modifier for the cache-sake (it cannot handle more than 1 cache on the same modifier-index) * no color on gui anymore * fixing non-power-of-2 resolutions (hopefully) * fixing select-deselect of domain drawing bug * fixing drawobject.c coding style (making Ton happy) ;-) HINT #1: If scons doesn't work -> cmakefiles are up-to-date, couldn't test scons (but i tried to mantain them, too) CODERS HINT #1: we really need a way to disable adding all modifiers through "Add Modifiers" dropdown! WARNING #1: before applying this commit, deactivate your SMOKE DOMAIN in your old files and save them then. You can open them then savely after that. WARNING #2: File and cache format of smoke can be changed, this is not final!
2009-08-20 00:33:59 +00:00
}
Smoke: * Less verbose * More OpenGL error messages (on blender -d)
2009-09-15 10:26:36 +00:00
Smoke: * cache for low res (deactivating high res for now) * new way of view3d rendering of smoke (no longer 3 axes) -using 3dtexture now (introduced into gpu/intern) * introducing LZO and LZMA libs into extern (makefiles missing for now) * reducing memory usage after simulating for the frame ended (freeing temporary buffers) * splitting smoke into 2 modifier for the cache-sake (it cannot handle more than 1 cache on the same modifier-index) * no color on gui anymore * fixing non-power-of-2 resolutions (hopefully) * fixing select-deselect of domain drawing bug * fixing drawobject.c coding style (making Ton happy) ;-) HINT #1: If scons doesn't work -> cmakefiles are up-to-date, couldn't test scons (but i tried to mantain them, too) CODERS HINT #1: we really need a way to disable adding all modifiers through "Add Modifiers" dropdown! WARNING #1: before applying this commit, deactivate your SMOKE DOMAIN in your old files and save them then. You can open them then savely after that. WARNING #2: File and cache format of smoke can be changed, this is not final!
2009-08-20 00:33:59 +00:00
glTexParameteri(GL_TEXTURE_3D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
glTexParameteri(GL_TEXTURE_3D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
Smoke: * Use GL_QUADS and GL_TRIANGLES instead of GL_POLYGON for faster drawing * Use variable count of slices
2009-10-20 15:51:25 +00:00
glTexParameteri(GL_TEXTURE_3D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
glTexParameteri(GL_TEXTURE_3D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
glTexParameteri(GL_TEXTURE_3D, GL_TEXTURE_WRAP_R, GL_CLAMP_TO_EDGE);
Smoke: * cache for low res (deactivating high res for now) * new way of view3d rendering of smoke (no longer 3 axes) -using 3dtexture now (introduced into gpu/intern) * introducing LZO and LZMA libs into extern (makefiles missing for now) * reducing memory usage after simulating for the frame ended (freeing temporary buffers) * splitting smoke into 2 modifier for the cache-sake (it cannot handle more than 1 cache on the same modifier-index) * no color on gui anymore * fixing non-power-of-2 resolutions (hopefully) * fixing select-deselect of domain drawing bug * fixing drawobject.c coding style (making Ton happy) ;-) HINT #1: If scons doesn't work -> cmakefiles are up-to-date, couldn't test scons (but i tried to mantain them, too) CODERS HINT #1: we really need a way to disable adding all modifiers through "Add Modifiers" dropdown! WARNING #1: before applying this commit, deactivate your SMOKE DOMAIN in your old files and save them then. You can open them then savely after that. WARNING #2: File and cache format of smoke can be changed, this is not final!
2009-08-20 00:33:59 +00:00
if (pixels)
MEM_freeN(pixels);
- fix for access past the buffer size (paint / sculpt used some 2d vecs as 3d) - remove redundant NULL checks on old code where it would crash if the result was NULL later on. - add some missing NULL checks.
2011-09-16 06:47:01 +00:00
GPU_texture_unbind(tex);
Smoke: * cache for low res (deactivating high res for now) * new way of view3d rendering of smoke (no longer 3 axes) -using 3dtexture now (introduced into gpu/intern) * introducing LZO and LZMA libs into extern (makefiles missing for now) * reducing memory usage after simulating for the frame ended (freeing temporary buffers) * splitting smoke into 2 modifier for the cache-sake (it cannot handle more than 1 cache on the same modifier-index) * no color on gui anymore * fixing non-power-of-2 resolutions (hopefully) * fixing select-deselect of domain drawing bug * fixing drawobject.c coding style (making Ton happy) ;-) HINT #1: If scons doesn't work -> cmakefiles are up-to-date, couldn't test scons (but i tried to mantain them, too) CODERS HINT #1: we really need a way to disable adding all modifiers through "Add Modifiers" dropdown! WARNING #1: before applying this commit, deactivate your SMOKE DOMAIN in your old files and save them then. You can open them then savely after that. WARNING #2: File and cache format of smoke can be changed, this is not final!
2009-08-20 00:33:59 +00:00
return tex;
}
Code cleanup: remove TRUE/FALSE & WITH_BOOL_COMPAT define
2014-04-01 11:34:00 +11:00
GPUTexture *GPU_texture_from_blender(Image *ima, ImageUser *iuser, bool is_data, double time, int mipmap)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
GPUTexture *tex;
GLint w, h, border, lastbindcode, bindcode;
glGetIntegerv(GL_TEXTURE_BINDING_2D, &lastbindcode);
GPU_update_image_time(ima, time);
Matcap support in 3D Viewport. Full log is here: http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.66/Usability#Matcap_in_3D_viewport Implementation notes: - Matcaps are an extension of Solid draw mode, and don't show in other drawmodes. (It's mostly intended to aid modeling/sculpt) - By design, Matcaps are a UI feature, and only stored locally for the UI itself, and won't affect rendering or materials. - Currently a set of 16 (GPL licensed) Matcaps have been compiled into Blender. It doesn't take memory or cpu time, until you use it. - Brush Icons and Matcaps use same code now, and only get generated/allocated on actually using it (instead of on startup). - The current set might get new or different images still, based on user feedback. - Matcap images are 512x512 pixels, so each image takes 1 Mb memory. Unused matcaps get freed immediately. The Matcap icon previews (128x128 pixels) stay in memory. - Loading own matcap image files will be added later. That needs design and code work to get it stable and memory-friendly. - The GLSL code uses the ID PreviewImage for matcaps. I tested it using the existing Material previews, which has its limits... especially for textured previews the normal-mapped matcap won't look good.
2013-01-22 11:18:41 +00:00
/* this binds a texture, so that's why to restore it with lastbindcode */
Code cleanup: remove TRUE/FALSE & WITH_BOOL_COMPAT define
2014-04-01 11:34:00 +11:00
bindcode = GPU_verify_image(ima, iuser, 0, 0, mipmap, is_data);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (ima->gputexture) {
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
ima->gputexture->bindcode = bindcode;
glBindTexture(GL_TEXTURE_2D, lastbindcode);
return ima->gputexture;
}
tex = MEM_callocN(sizeof(GPUTexture), "GPUTexture");
tex->bindcode = bindcode;
tex->number = -1;
tex->refcount = 1;
tex->target = GL_TEXTURE_2D;
tex->fromblender = 1;
ima->gputexture= tex;
if (!glIsTexture(tex->bindcode)) {
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_ASSERT_NO_GL_ERRORS("Blender Texture Not Loaded");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
else {
glBindTexture(GL_TEXTURE_2D, tex->bindcode);
glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_WIDTH, &w);
glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_HEIGHT, &h);
glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_BORDER, &border);
Alternate fix for offscreen render w/o npot Store the original texture size, since its needed for reading pixels and passing to glViewport.
2015-09-14 13:18:43 +10:00
tex->w = tex->w_orig = w - border;
tex->h = tex->h_orig = h - border;
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
glBindTexture(GL_TEXTURE_2D, lastbindcode);
return tex;
}
Matcap support in 3D Viewport. Full log is here: http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.66/Usability#Matcap_in_3D_viewport Implementation notes: - Matcaps are an extension of Solid draw mode, and don't show in other drawmodes. (It's mostly intended to aid modeling/sculpt) - By design, Matcaps are a UI feature, and only stored locally for the UI itself, and won't affect rendering or materials. - Currently a set of 16 (GPL licensed) Matcaps have been compiled into Blender. It doesn't take memory or cpu time, until you use it. - Brush Icons and Matcaps use same code now, and only get generated/allocated on actually using it (instead of on startup). - The current set might get new or different images still, based on user feedback. - Matcap images are 512x512 pixels, so each image takes 1 Mb memory. Unused matcaps get freed immediately. The Matcap icon previews (128x128 pixels) stay in memory. - Loading own matcap image files will be added later. That needs design and code work to get it stable and memory-friendly. - The GLSL code uses the ID PreviewImage for matcaps. I tested it using the existing Material previews, which has its limits... especially for textured previews the normal-mapped matcap won't look good.
2013-01-22 11:18:41 +00:00
GPUTexture *GPU_texture_from_preview(PreviewImage *prv, int mipmap)
{
GPUTexture *tex = prv->gputexture[0];
GLint w, h, lastbindcode;
GLuint bindcode = 0;
glGetIntegerv(GL_TEXTURE_BINDING_2D, &lastbindcode);
if (tex)
bindcode = tex->bindcode;
/* this binds a texture, so that's why to restore it */
if (bindcode == 0) {
GPU_create_gl_tex(&bindcode, prv->rect[0], NULL, prv->w[0], prv->h[0], mipmap, 0, NULL);
}
if (tex) {
tex->bindcode = bindcode;
glBindTexture(GL_TEXTURE_2D, lastbindcode);
return tex;
}
tex = MEM_callocN(sizeof(GPUTexture), "GPUTexture");
tex->bindcode = bindcode;
tex->number = -1;
tex->refcount = 1;
tex->target = GL_TEXTURE_2D;
Cleanup: style
2015-06-24 05:13:43 +10:00
prv->gputexture[0] = tex;
Matcap support in 3D Viewport. Full log is here: http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.66/Usability#Matcap_in_3D_viewport Implementation notes: - Matcaps are an extension of Solid draw mode, and don't show in other drawmodes. (It's mostly intended to aid modeling/sculpt) - By design, Matcaps are a UI feature, and only stored locally for the UI itself, and won't affect rendering or materials. - Currently a set of 16 (GPL licensed) Matcaps have been compiled into Blender. It doesn't take memory or cpu time, until you use it. - Brush Icons and Matcaps use same code now, and only get generated/allocated on actually using it (instead of on startup). - The current set might get new or different images still, based on user feedback. - Matcap images are 512x512 pixels, so each image takes 1 Mb memory. Unused matcaps get freed immediately. The Matcap icon previews (128x128 pixels) stay in memory. - Loading own matcap image files will be added later. That needs design and code work to get it stable and memory-friendly. - The GLSL code uses the ID PreviewImage for matcaps. I tested it using the existing Material previews, which has its limits... especially for textured previews the normal-mapped matcap won't look good.
2013-01-22 11:18:41 +00:00
if (!glIsTexture(tex->bindcode)) {
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_ASSERT_NO_GL_ERRORS("Blender Texture Not Loaded");
Matcap support in 3D Viewport. Full log is here: http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.66/Usability#Matcap_in_3D_viewport Implementation notes: - Matcaps are an extension of Solid draw mode, and don't show in other drawmodes. (It's mostly intended to aid modeling/sculpt) - By design, Matcaps are a UI feature, and only stored locally for the UI itself, and won't affect rendering or materials. - Currently a set of 16 (GPL licensed) Matcaps have been compiled into Blender. It doesn't take memory or cpu time, until you use it. - Brush Icons and Matcaps use same code now, and only get generated/allocated on actually using it (instead of on startup). - The current set might get new or different images still, based on user feedback. - Matcap images are 512x512 pixels, so each image takes 1 Mb memory. Unused matcaps get freed immediately. The Matcap icon previews (128x128 pixels) stay in memory. - Loading own matcap image files will be added later. That needs design and code work to get it stable and memory-friendly. - The GLSL code uses the ID PreviewImage for matcaps. I tested it using the existing Material previews, which has its limits... especially for textured previews the normal-mapped matcap won't look good.
2013-01-22 11:18:41 +00:00
}
else {
glBindTexture(GL_TEXTURE_2D, tex->bindcode);
glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_WIDTH, &w);
glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_HEIGHT, &h);
Alternate fix for offscreen render w/o npot Store the original texture size, since its needed for reading pixels and passing to glViewport.
2015-09-14 13:18:43 +10:00
tex->w = tex->w_orig = w;
tex->h = tex->h_orig = h;
Matcap support in 3D Viewport. Full log is here: http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.66/Usability#Matcap_in_3D_viewport Implementation notes: - Matcaps are an extension of Solid draw mode, and don't show in other drawmodes. (It's mostly intended to aid modeling/sculpt) - By design, Matcaps are a UI feature, and only stored locally for the UI itself, and won't affect rendering or materials. - Currently a set of 16 (GPL licensed) Matcaps have been compiled into Blender. It doesn't take memory or cpu time, until you use it. - Brush Icons and Matcaps use same code now, and only get generated/allocated on actually using it (instead of on startup). - The current set might get new or different images still, based on user feedback. - Matcap images are 512x512 pixels, so each image takes 1 Mb memory. Unused matcaps get freed immediately. The Matcap icon previews (128x128 pixels) stay in memory. - Loading own matcap image files will be added later. That needs design and code work to get it stable and memory-friendly. - The GLSL code uses the ID PreviewImage for matcaps. I tested it using the existing Material previews, which has its limits... especially for textured previews the normal-mapped matcap won't look good.
2013-01-22 11:18:41 +00:00
}
glBindTexture(GL_TEXTURE_2D, lastbindcode);
return tex;
}
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
GPUTexture *GPU_texture_create_1D(int w, const float *fpixels, char err_out[256])
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
GPUTexture *tex = GPU_texture_create_nD(w, 1, 1, fpixels, 0, GPU_HDR_NONE, 4, err_out);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
if (tex)
GPU_texture_unbind(tex);
return tex;
}
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
GPUTexture *GPU_texture_create_2D(int w, int h, const float *fpixels, GPUHDRType hdr, char err_out[256])
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
GPUTexture *tex = GPU_texture_create_nD(w, h, 2, fpixels, 0, hdr, 4, err_out);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
if (tex)
GPU_texture_unbind(tex);
return tex;
}
bug [#26329] Project Paint not working we cant ensure that a requested buffer can be allocated so report opengl errors when failing to allocate the buffer (rather then printing to console). this is common enough and generic error isn't too helpful to users.
2011-03-06 23:12:12 +00:00
GPUTexture *GPU_texture_create_depth(int w, int h, char err_out[256])
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
GPUTexture *tex = GPU_texture_create_nD(w, h, 2, NULL, 1, GPU_HDR_NONE, 1, err_out);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
if (tex)
GPU_texture_unbind(tex);
return tex;
}
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
/**
* A shadow map for VSM needs two components (depth and depth^2)
*/
GPUTexture *GPU_texture_create_vsm_shadow_map(int size, char err_out[256])
{
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
GPUTexture *tex = GPU_texture_create_nD(size, size, 2, NULL, 0, GPU_HDR_FULL_FLOAT, 2, err_out);
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
if (tex) {
/* Now we tweak some of the settings */
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
GPU_texture_unbind(tex);
}
return tex;
}
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
GPUTexture *GPU_texture_create_2D_procedural(int w, int h, const float *pixels, bool repeat, char err_out[256])
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
{
GPUTexture *tex = GPU_texture_create_nD(w, h, 2, pixels, 0, GPU_HDR_HALF_FLOAT, 2, err_out);
if (tex) {
/* Now we tweak some of the settings */
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
if (repeat) {
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_REPEAT);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_REPEAT);
}
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
GPU_texture_unbind(tex);
}
return tex;
}
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
GPUTexture *GPU_texture_create_1D_procedural(int w, const float *pixels, char err_out[256])
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
{
GPUTexture *tex = GPU_texture_create_nD(w, 0, 1, pixels, 0, GPU_HDR_HALF_FLOAT, 2, err_out);
if (tex) {
/* Now we tweak some of the settings */
glTexParameteri(GL_TEXTURE_1D, GL_TEXTURE_WRAP_S, GL_REPEAT);
glTexParameteri(GL_TEXTURE_1D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
glTexParameteri(GL_TEXTURE_1D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
GPU_texture_unbind(tex);
}
return tex;
}
Fix T40498 invalid textures flickering. Issue here is most likely sampler uniforms and textures not being updated properly when zero binding is created. Solution for now is to allow zero binding but when this happens use sexy pink invalid texture instead :p.
2014-06-10 01:36:53 +03:00
void GPU_invalid_tex_init(void)
{
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
const float color[4] = {1.0f, 0.0f, 1.0f, 1.0f};
Fix T40498 invalid textures flickering. Issue here is most likely sampler uniforms and textures not being updated properly when zero binding is created. Solution for now is to allow zero binding but when this happens use sexy pink invalid texture instead :p.
2014-06-10 01:36:53 +03:00
GG.invalid_tex_1D = GPU_texture_create_1D(1, color, NULL);
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
GG.invalid_tex_2D = GPU_texture_create_2D(1, 1, color, GPU_HDR_NONE, NULL);
Fix T40498 invalid textures flickering. Issue here is most likely sampler uniforms and textures not being updated properly when zero binding is created. Solution for now is to allow zero binding but when this happens use sexy pink invalid texture instead :p.
2014-06-10 01:36:53 +03:00
GG.invalid_tex_3D = GPU_texture_create_3D(1, 1, 1, 4, color);
}
void GPU_invalid_tex_bind(int mode)
{
Quiet double promotion warning & ws edit
2014-06-13 02:22:40 +10:00
switch (mode) {
Fix T40498 invalid textures flickering. Issue here is most likely sampler uniforms and textures not being updated properly when zero binding is created. Solution for now is to allow zero binding but when this happens use sexy pink invalid texture instead :p.
2014-06-10 01:36:53 +03:00
case GL_TEXTURE_1D:
glBindTexture(GL_TEXTURE_1D, GG.invalid_tex_1D->bindcode);
break;
case GL_TEXTURE_2D:
glBindTexture(GL_TEXTURE_2D, GG.invalid_tex_2D->bindcode);
break;
case GL_TEXTURE_3D:
glBindTexture(GL_TEXTURE_3D, GG.invalid_tex_3D->bindcode);
break;
}
}
void GPU_invalid_tex_free(void)
{
Fix T42049: Crash exiting /w GL1.1
2014-10-02 16:21:26 +02:00
if (GG.invalid_tex_1D)
GPU_texture_free(GG.invalid_tex_1D);
if (GG.invalid_tex_2D)
GPU_texture_free(GG.invalid_tex_2D);
if (GG.invalid_tex_3D)
GPU_texture_free(GG.invalid_tex_3D);
Fix T40498 invalid textures flickering. Issue here is most likely sampler uniforms and textures not being updated properly when zero binding is created. Solution for now is to allow zero binding but when this happens use sexy pink invalid texture instead :p.
2014-06-10 01:36:53 +03:00
}
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
void GPU_texture_bind(GPUTexture *tex, int number)
{
GLenum arbnumber;
if (number >= GG.maxtextures) {
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
fprintf(stderr, "Not enough texture slots.");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
return;
}
GPU: code cleanup, no fundamental changes
2014-12-03 17:26:43 -02:00
if ((G.debug & G_DEBUG)) {
if (tex->fb && tex->fb->object == GG.currentfb) {
GPU framebuffer/texture API: Warn when binding a texture that is also attached to a framebuffer or vice versa. might be more correct to just handle the case and unbind here.
2014-11-18 12:12:28 +01:00
fprintf(stderr, "Feedback loop warning!: Attempting to bind texture attached to current framebuffer!\n");
}
}
if (number < 0)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
return;
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_ASSERT_NO_GL_ERRORS("Pre Texture Bind");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
arbnumber = (GLenum)((GLuint)GL_TEXTURE0_ARB + number);
if (number != 0) glActiveTextureARB(arbnumber);
Fix T40498 invalid textures flickering. Issue here is most likely sampler uniforms and textures not being updated properly when zero binding is created. Solution for now is to allow zero binding but when this happens use sexy pink invalid texture instead :p.
2014-06-10 01:36:53 +03:00
if (tex->bindcode != 0) {
glBindTexture(tex->target, tex->bindcode);
}
else
GPU_invalid_tex_bind(tex->target);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
glEnable(tex->target);
if (number != 0) glActiveTextureARB(GL_TEXTURE0_ARB);
tex->number = number;
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_ASSERT_NO_GL_ERRORS("Post Texture Bind");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
void GPU_texture_unbind(GPUTexture *tex)
{
GLenum arbnumber;
if (tex->number >= GG.maxtextures) {
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
fprintf(stderr, "Not enough texture slots.");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
return;
}
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (tex->number == -1)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
return;
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_ASSERT_NO_GL_ERRORS("Pre Texture Unbind");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
arbnumber = (GLenum)((GLuint)GL_TEXTURE0_ARB + tex->number);
if (tex->number != 0) glActiveTextureARB(arbnumber);
glBindTexture(tex->target, 0);
glDisable(tex->target);
if (tex->number != 0) glActiveTextureARB(GL_TEXTURE0_ARB);
tex->number = -1;
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_ASSERT_NO_GL_ERRORS("Post Texture Unbind");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
void GPU_texture_filter_mode(GPUTexture *tex, bool compare, bool use_filter)
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
{
GLenum arbnumber;
if (tex->number >= GG.maxtextures) {
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
fprintf(stderr, "Not enough texture slots.");
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
return;
}
if (tex->number == -1)
return;
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_ASSERT_NO_GL_ERRORS("Pre Texture Unbind");
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
arbnumber = (GLenum)((GLuint)GL_TEXTURE0_ARB + tex->number);
if (tex->number != 0) glActiveTextureARB(arbnumber);
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
if (tex->depth) {
if (compare)
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_COMPARE_MODE, GL_COMPARE_R_TO_TEXTURE);
else
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_COMPARE_MODE, GL_NONE);
}
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
if (use_filter) {
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
}
else {
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
}
if (tex->number != 0) glActiveTextureARB(GL_TEXTURE0_ARB);
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_ASSERT_NO_GL_ERRORS("Post Texture Unbind");
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
}
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
void GPU_texture_free(GPUTexture *tex)
{
tex->refcount--;
if (tex->refcount < 0)
fprintf(stderr, "GPUTexture: negative refcount\n");
if (tex->refcount == 0) {
if (tex->fb)
GPU framebuffer API: Allow binding a texture to a different texture attachment than the first. Also fix a number error in seperable gaussian blur shader.
2014-11-18 11:56:23 +01:00
GPU_framebuffer_texture_detach(tex);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
if (tex->bindcode && !tex->fromblender)
glDeleteTextures(1, &tex->bindcode);
MEM_freeN(tex);
}
}
void GPU_texture_ref(GPUTexture *tex)
{
tex->refcount++;
}
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
int GPU_texture_target(const GPUTexture *tex)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
return tex->target;
}
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
int GPU_texture_opengl_width(const GPUTexture *tex)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
return tex->w;
}
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
int GPU_texture_opengl_height(const GPUTexture *tex)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
return tex->h;
}
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
int GPU_texture_opengl_bindcode(const GPUTexture *tex)
GPU: add gpu python module with export_shader() function to export GLSL shader. shader = gpu.export_shader(scene,material) Returns the GLSL shader that blender generates to produce the visual effect of material in scene for the purpose of reusing the shader in an external engine. This function is meant to be used in a material exporter so that the GLSL shader can be exported entirely. The return value is a dictionary containing the shader source code and all associated data. The full documentation is under sphinx. Warning: there has been an API between the patch and this commit: uniform['lamp'] and uniform['image'] now return python reference to ID block instead of ID name as before. The X3D exporter that uses this function must be adapted.
2011-09-09 11:55:38 +00:00
{
return tex->bindcode;
}
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
GPUFrameBuffer *GPU_texture_framebuffer(GPUTexture *tex)
{
return tex->fb;
}
/* GPUFrameBuffer */
warning cleanup. - fix mistake with grease pencil UI (&& was intended but & used). - use (void) rather then () across _all_ blenders code. - a few minor edits, don't shadow stack variables in roll calculation & avoid running memset() for VBO vertex map.
2011-02-13 15:02:21 +00:00
GPUFrameBuffer *GPU_framebuffer_create(void)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
GPUFrameBuffer *fb;
if (!GLEW_EXT_framebuffer_object)
return NULL;
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
fb = MEM_callocN(sizeof(GPUFrameBuffer), "GPUFrameBuffer");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
glGenFramebuffersEXT(1, &fb->object);
if (!fb->object) {
fprintf(stderr, "GPUFFrameBuffer: framebuffer gen failed. %d\n",
(int)glGetError());
GPU_framebuffer_free(fb);
return NULL;
}
Fix framebuffer completeness being broken after last framebuffer cleanup commits. Basically, we don't set a draw buffer until draw time comes. Also add explicit validation function to validate after all textures have been attached (could be done automatically at bind time too probably, but left out for now)
2014-11-25 16:16:50 +01:00
/* make sure no read buffer is enabled, so completeness check will not fail. We set those at binding time */
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, fb->object);
glReadBuffer(GL_NONE);
glDrawBuffer(GL_NONE);
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, 0);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
return fb;
}
GPU framebuffer API: Allow binding a texture to a different texture attachment than the first. Also fix a number error in seperable gaussian blur shader.
2014-11-18 11:56:23 +01:00
int GPU_framebuffer_texture_attach(GPUFrameBuffer *fb, GPUTexture *tex, int slot, char err_out[256])
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
GLenum attachment;
Fix #33371: blender freezing in material draw mode. When FBO failed in a particular way it could cause the opengl draw buffer to be set wrong, effectively disabling all opengl drawing. The FBO error was caused by cycles GLSL materials with no nodes that would still use blender internal materials, which caused issues with lamp shadow buffers FBO. This also fixes a GLSL refresh issue when switching render engines.
2012-12-03 08:31:16 +00:00
GLenum error;
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
GPU framebuffer API: Allow binding a texture to a different texture attachment than the first. Also fix a number error in seperable gaussian blur shader.
2014-11-18 11:56:23 +01:00
if (slot >= GPU_FB_MAX_SLOTS) {
fprintf(stderr, "Attaching to index %d framebuffer slot unsupported in blender use at most %d\n", slot, GPU_FB_MAX_SLOTS);
return 0;
}
GPU: code cleanup, no fundamental changes
2014-12-03 17:26:43 -02:00
if ((G.debug & G_DEBUG)) {
if (tex->number != -1) {
fprintf(stderr, "Feedback loop warning!: Attempting to attach texture to framebuffer while still bound to texture unit for drawing!");
}
GPU framebuffer/texture API: Warn when binding a texture that is also attached to a framebuffer or vice versa. might be more correct to just handle the case and unbind here.
2014-11-18 12:12:28 +01:00
}
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (tex->depth)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
attachment = GL_DEPTH_ATTACHMENT_EXT;
else
GPU framebuffer API: Allow binding a texture to a different texture attachment than the first. Also fix a number error in seperable gaussian blur shader.
2014-11-18 11:56:23 +01:00
attachment = GL_COLOR_ATTACHMENT0_EXT + slot;
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, fb->object);
GG.currentfb = fb->object;
Fix T41314: OpenGL error when using Cycles engine
2014-08-18 13:54:29 +06:00
/* Clean glError buffer. */
while (glGetError() != GL_NO_ERROR) {}
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
glFramebufferTexture2DEXT(GL_FRAMEBUFFER_EXT, attachment,
tex->target, tex->bindcode, 0);
Fix #33371: blender freezing in material draw mode. When FBO failed in a particular way it could cause the opengl draw buffer to be set wrong, effectively disabling all opengl drawing. The FBO error was caused by cycles GLSL materials with no nodes that would still use blender internal materials, which caused issues with lamp shadow buffers FBO. This also fixes a GLSL refresh issue when switching render engines.
2012-12-03 08:31:16 +00:00
error = glGetError();
if (error == GL_INVALID_OPERATION) {
GPU_framebuffer_restore();
GPU_print_framebuffer_error(error, err_out);
return 0;
}
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (tex->depth)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
fb->depthtex = tex;
else
GPU framebuffer API: Allow binding a texture to a different texture attachment than the first. Also fix a number error in seperable gaussian blur shader.
2014-11-18 11:56:23 +01:00
fb->colortex[slot] = tex;
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
tex->fb= fb;
GPU framebuffer API: Allow binding a texture to a different texture attachment than the first. Also fix a number error in seperable gaussian blur shader.
2014-11-18 11:56:23 +01:00
tex->fb_attachment = slot;
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
return 1;
}
GPU framebuffer API: Allow binding a texture to a different texture attachment than the first. Also fix a number error in seperable gaussian blur shader.
2014-11-18 11:56:23 +01:00
void GPU_framebuffer_texture_detach(GPUTexture *tex)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
GLenum attachment;
GPU framebuffer API: Allow binding a texture to a different texture attachment than the first. Also fix a number error in seperable gaussian blur shader.
2014-11-18 11:56:23 +01:00
GPUFrameBuffer *fb;
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
GPU framebuffer/texture API: Warn when binding a texture that is also attached to a framebuffer or vice versa. might be more correct to just handle the case and unbind here.
2014-11-18 12:12:28 +01:00
if (!tex->fb)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
return;
GPU framebuffer API: Allow binding a texture to a different texture attachment than the first. Also fix a number error in seperable gaussian blur shader.
2014-11-18 11:56:23 +01:00
fb = tex->fb;
if (GG.currentfb != fb->object) {
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, fb->object);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
GG.currentfb = tex->fb->object;
}
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (tex->depth) {
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
fb->depthtex = NULL;
attachment = GL_DEPTH_ATTACHMENT_EXT;
}
else {
GPU framebuffer API: Allow binding a texture to a different texture attachment than the first. Also fix a number error in seperable gaussian blur shader.
2014-11-18 11:56:23 +01:00
BLI_assert(fb->colortex[tex->fb_attachment] == tex);
fb->colortex[tex->fb_attachment] = NULL;
attachment = GL_COLOR_ATTACHMENT0_EXT + tex->fb_attachment;
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
glFramebufferTexture2DEXT(GL_FRAMEBUFFER_EXT, attachment, tex->target, 0, 0);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
tex->fb = NULL;
GPU framebuffer API: Allow binding a texture to a different texture attachment than the first. Also fix a number error in seperable gaussian blur shader.
2014-11-18 11:56:23 +01:00
tex->fb_attachment = -1;
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
GPUFramebuffer API cleanup: * read buffers are set at texture binding time * change naming when setting a texture as framebuffer * add function to set slot of framebuffer as current target instead of texture. * Binding a buffer reuses the dimensions of the texture at bind time (can use viewport to set to arbitrary range later) * Removed offscreen buffer width/height, use the generated texture dimensions instead. Those were supposed to be checked to see if generated texture had the requested size but were never actually changed to the texture dimensions (and it's redundant to store twice).
2014-11-18 12:37:55 +01:00
void GPU_texture_bind_as_framebuffer(GPUTexture *tex)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
GPUFramebuffer API cleanup: * read buffers are set at texture binding time * change naming when setting a texture as framebuffer * add function to set slot of framebuffer as current target instead of texture. * Binding a buffer reuses the dimensions of the texture at bind time (can use viewport to set to arbitrary range later) * Removed offscreen buffer width/height, use the generated texture dimensions instead. Those were supposed to be checked to see if generated texture had the requested size but were never actually changed to the texture dimensions (and it's redundant to store twice).
2014-11-18 12:37:55 +01:00
if (!tex->fb) {
fprintf(stderr, "Error, texture not bound to framebuffer!");
return;
}
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
/* push attributes */
Fix #34997: when starting the game engine in one window and switching to a second window, the game would stop drawing in the first and mess up the OpenGL state of the second. Also fixes glPushAttrib/glPopAttrib getting out of sync in some cases.
2013-04-18 16:28:39 +00:00
glPushAttrib(GL_ENABLE_BIT | GL_VIEWPORT_BIT);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
glDisable(GL_SCISSOR_TEST);
/* bind framebuffer */
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, tex->fb->object);
Fix T42917 shadow maps not working on ATIs. This is yet another issue with framebuffers. There are two issues: We need the framebuffer fully bound to check for completeness and when we bind a depth texture as frame buffer we need to disable read/write.
2014-12-16 21:52:55 +01:00
if (tex->depth) {
glDrawBuffer(GL_NONE);
glReadBuffer(GL_NONE);
}
else {
/* last bound prevails here, better allow explicit control here too */
glDrawBuffer(GL_COLOR_ATTACHMENT0_EXT + tex->fb_attachment);
glReadBuffer(GL_COLOR_ATTACHMENT0_EXT + tex->fb_attachment);
}
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
/* push matrices and set default viewport and matrix */
Alternate fix for offscreen render w/o npot Store the original texture size, since its needed for reading pixels and passing to glViewport.
2015-09-14 13:18:43 +10:00
glViewport(0, 0, tex->w_orig, tex->h_orig);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
GG.currentfb = tex->fb->object;
glMatrixMode(GL_PROJECTION);
glPushMatrix();
glMatrixMode(GL_MODELVIEW);
glPushMatrix();
}
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
void GPU_framebuffer_slots_bind(GPUFrameBuffer *fb, int slot)
GPUFramebuffer API cleanup: * read buffers are set at texture binding time * change naming when setting a texture as framebuffer * add function to set slot of framebuffer as current target instead of texture. * Binding a buffer reuses the dimensions of the texture at bind time (can use viewport to set to arbitrary range later) * Removed offscreen buffer width/height, use the generated texture dimensions instead. Those were supposed to be checked to see if generated texture had the requested size but were never actually changed to the texture dimensions (and it's redundant to store twice).
2014-11-18 12:37:55 +01:00
{
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
int numslots = 0, i;
GLenum attachments[4];
GPUFramebuffer API cleanup: * read buffers are set at texture binding time * change naming when setting a texture as framebuffer * add function to set slot of framebuffer as current target instead of texture. * Binding a buffer reuses the dimensions of the texture at bind time (can use viewport to set to arbitrary range later) * Removed offscreen buffer width/height, use the generated texture dimensions instead. Those were supposed to be checked to see if generated texture had the requested size but were never actually changed to the texture dimensions (and it's redundant to store twice).
2014-11-18 12:37:55 +01:00
if (!fb->colortex[slot]) {
fprintf(stderr, "Error, framebuffer slot empty!");
return;
}
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
for (i = 0; i < 4; i++) {
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
if (fb->colortex[i]) {
attachments[numslots] = GL_COLOR_ATTACHMENT0_EXT + i;
numslots++;
}
}
GPUFramebuffer API cleanup: * read buffers are set at texture binding time * change naming when setting a texture as framebuffer * add function to set slot of framebuffer as current target instead of texture. * Binding a buffer reuses the dimensions of the texture at bind time (can use viewport to set to arbitrary range later) * Removed offscreen buffer width/height, use the generated texture dimensions instead. Those were supposed to be checked to see if generated texture had the requested size but were never actually changed to the texture dimensions (and it's redundant to store twice).
2014-11-18 12:37:55 +01:00
/* push attributes */
glPushAttrib(GL_ENABLE_BIT | GL_VIEWPORT_BIT);
glDisable(GL_SCISSOR_TEST);
/* bind framebuffer */
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, fb->object);
/* last bound prevails here, better allow explicit control here too */
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
glDrawBuffers(numslots, attachments);
GPUFramebuffer API cleanup: * read buffers are set at texture binding time * change naming when setting a texture as framebuffer * add function to set slot of framebuffer as current target instead of texture. * Binding a buffer reuses the dimensions of the texture at bind time (can use viewport to set to arbitrary range later) * Removed offscreen buffer width/height, use the generated texture dimensions instead. Those were supposed to be checked to see if generated texture had the requested size but were never actually changed to the texture dimensions (and it's redundant to store twice).
2014-11-18 12:37:55 +01:00
glReadBuffer(GL_COLOR_ATTACHMENT0_EXT + slot);
/* push matrices and set default viewport and matrix */
Alternate fix for offscreen render w/o npot Store the original texture size, since its needed for reading pixels and passing to glViewport.
2015-09-14 13:18:43 +10:00
glViewport(0, 0, fb->colortex[slot]->w_orig, fb->colortex[slot]->h_orig);
GPUFramebuffer API cleanup: * read buffers are set at texture binding time * change naming when setting a texture as framebuffer * add function to set slot of framebuffer as current target instead of texture. * Binding a buffer reuses the dimensions of the texture at bind time (can use viewport to set to arbitrary range later) * Removed offscreen buffer width/height, use the generated texture dimensions instead. Those were supposed to be checked to see if generated texture had the requested size but were never actually changed to the texture dimensions (and it's redundant to store twice).
2014-11-18 12:37:55 +01:00
GG.currentfb = fb->object;
glMatrixMode(GL_PROJECTION);
glPushMatrix();
glMatrixMode(GL_MODELVIEW);
glPushMatrix();
}
remove/tag unused args for view*.c, gpu*.c & image*.c
2010-10-15 12:29:02 +00:00
void GPU_framebuffer_texture_unbind(GPUFrameBuffer *UNUSED(fb), GPUTexture *UNUSED(tex))
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
/* restore matrix */
glMatrixMode(GL_PROJECTION);
glPopMatrix();
glMatrixMode(GL_MODELVIEW);
glPopMatrix();
/* restore attributes */
glPopAttrib();
}
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
void GPU_framebuffer_bind_no_save(GPUFrameBuffer *fb, int slot)
{
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, fb->object);
/* last bound prevails here, better allow explicit control here too */
glDrawBuffer(GL_COLOR_ATTACHMENT0_EXT + slot);
glReadBuffer(GL_COLOR_ATTACHMENT0_EXT + slot);
/* push matrices and set default viewport and matrix */
Alternate fix for offscreen render w/o npot Store the original texture size, since its needed for reading pixels and passing to glViewport.
2015-09-14 13:18:43 +10:00
glViewport(0, 0, fb->colortex[slot]->w_orig, fb->colortex[slot]->h_orig);
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
GG.currentfb = fb->object;
GG.currentfb = fb->object;
}
Fix framebuffer completeness being broken after last framebuffer cleanup commits. Basically, we don't set a draw buffer until draw time comes. Also add explicit validation function to validate after all textures have been attached (could be done automatically at bind time too probably, but left out for now)
2014-11-25 16:16:50 +01:00
bool GPU_framebuffer_check_valid(GPUFrameBuffer *fb, char err_out[256])
{
GLenum status;
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, fb->object);
GG.currentfb = fb->object;
/* Clean glError buffer. */
while (glGetError() != GL_NO_ERROR) {}
status = glCheckFramebufferStatusEXT(GL_FRAMEBUFFER_EXT);
if (status != GL_FRAMEBUFFER_COMPLETE_EXT) {
GPU_framebuffer_restore();
GPU_print_framebuffer_error(status, err_out);
return false;
}
return true;
}
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
void GPU_framebuffer_free(GPUFrameBuffer *fb)
{
GPU framebuffer API: Allow binding a texture to a different texture attachment than the first. Also fix a number error in seperable gaussian blur shader.
2014-11-18 11:56:23 +01:00
int i;
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (fb->depthtex)
GPU framebuffer API: Allow binding a texture to a different texture attachment than the first. Also fix a number error in seperable gaussian blur shader.
2014-11-18 11:56:23 +01:00
GPU_framebuffer_texture_detach(fb->depthtex);
for (i = 0; i < GPU_FB_MAX_SLOTS; i++) {
if (fb->colortex[i]) {
GPU_framebuffer_texture_detach(fb->colortex[i]);
}
}
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (fb->object) {
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
glDeleteFramebuffersEXT(1, &fb->object);
if (GG.currentfb == fb->object) {
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, 0);
GG.currentfb = 0;
}
}
MEM_freeN(fb);
}
warning cleanup. - fix mistake with grease pencil UI (&& was intended but & used). - use (void) rather then () across _all_ blenders code. - a few minor edits, don't shadow stack variables in roll calculation & avoid running memset() for VBO vertex map.
2011-02-13 15:02:21 +00:00
void GPU_framebuffer_restore(void)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
if (GG.currentfb != 0) {
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, 0);
GG.currentfb = 0;
}
}
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
void GPU_framebuffer_blur(GPUFrameBuffer *fb, GPUTexture *tex, GPUFrameBuffer *blurfb, GPUTexture *blurtex)
{
Alternate fix for offscreen render w/o npot Store the original texture size, since its needed for reading pixels and passing to glViewport.
2015-09-14 13:18:43 +10:00
const float scaleh[2] = {1.0f / blurtex->w_orig, 0.0f};
const float scalev[2] = {0.0f, 1.0f / tex->h_orig};
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
GPUShader *blur_shader = GPU_shader_get_builtin_shader(GPU_SHADER_SEP_GAUSSIAN_BLUR);
int scale_uniform, texture_source_uniform;
if (!blur_shader)
return;
scale_uniform = GPU_shader_get_uniform(blur_shader, "ScaleU");
texture_source_uniform = GPU_shader_get_uniform(blur_shader, "textureSource");
/* Blurring horizontally */
/* We do the bind ourselves rather than using GPU_framebuffer_texture_bind() to avoid
style cleanup: block comments
2012-05-16 23:37:23 +00:00
* pushing unnecessary matrices onto the OpenGL stack. */
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, blurfb->object);
Fix T42807, variance shadow map broken afterFBO changes. This code was a bit do-it-yourself instead of relying on the API. Should be OK now.
2014-12-05 16:01:08 +01:00
glDrawBuffer(GL_COLOR_ATTACHMENT0);
/* avoid warnings from texture binding */
GG.currentfb = blurfb->object;
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
GPU_shader_bind(blur_shader);
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
GPU_shader_uniform_vector(blur_shader, scale_uniform, 2, 1, scaleh);
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
GPU_shader_uniform_texture(blur_shader, texture_source_uniform, tex);
Alternate fix for offscreen render w/o npot Store the original texture size, since its needed for reading pixels and passing to glViewport.
2015-09-14 13:18:43 +10:00
glViewport(0, 0, blurtex->w_orig, blurtex->h_orig);
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
/* Peparing to draw quad */
glMatrixMode(GL_MODELVIEW);
glLoadIdentity();
glMatrixMode(GL_TEXTURE);
glLoadIdentity();
glMatrixMode(GL_PROJECTION);
glLoadIdentity();
BGE fix [#35472] Sun variance shadows does not work in game engine The game engine had depth testing enabled when blurring the shadows which caused the blurring to not work properly.
2013-07-23 00:52:49 +00:00
glDisable(GL_DEPTH_TEST);
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
GPU_texture_bind(tex, 0);
/* Drawing quad */
glBegin(GL_QUADS);
style cleanup: whitespace, braces
2012-05-01 20:08:23 +00:00
glTexCoord2d(0, 0); glVertex2f(1, 1);
glTexCoord2d(1, 0); glVertex2f(-1, 1);
glTexCoord2d(1, 1); glVertex2f(-1, -1);
glTexCoord2d(0, 1); glVertex2f(1, -1);
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
glEnd();
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
/* Blurring vertically */
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, fb->object);
Fix T42807, variance shadow map broken afterFBO changes. This code was a bit do-it-yourself instead of relying on the API. Should be OK now.
2014-12-05 16:01:08 +01:00
glDrawBuffer(GL_COLOR_ATTACHMENT0);
GG.currentfb = fb->object;
Alternate fix for offscreen render w/o npot Store the original texture size, since its needed for reading pixels and passing to glViewport.
2015-09-14 13:18:43 +10:00
glViewport(0, 0, tex->w_orig, tex->h_orig);
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
GPU_shader_uniform_vector(blur_shader, scale_uniform, 2, 1, scalev);
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
GPU_shader_uniform_texture(blur_shader, texture_source_uniform, blurtex);
GPU_texture_bind(blurtex, 0);
style cleanup: whitespace, braces
2012-05-01 20:08:23 +00:00
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
glBegin(GL_QUADS);
style cleanup: whitespace, braces
2012-05-01 20:08:23 +00:00
glTexCoord2d(0, 0); glVertex2f(1, 1);
glTexCoord2d(1, 0); glVertex2f(-1, 1);
glTexCoord2d(1, 1); glVertex2f(-1, -1);
glTexCoord2d(0, 1); glVertex2f(1, -1);
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
glEnd();
style cleanup: whitespace, braces
2012-05-01 20:08:23 +00:00
OpenGL: implemenation of fixed function lighting as per pixel GLSL shaders. The code is still unused, but the intention is to use this to solve the double sided lighting problem on NVidia, and to make the materials work on OpenGL ES 2.0 eventually. The code works and matches the fixed function lighting pretty much exactly, but still needs optimizations. The actual integration in object draw will be committed later when more fixing & testing, there's lots of different combinations and unclear OpenGL state here.
2013-02-26 00:49:42 +00:00
GPU_shader_unbind();
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
}
OpenGL Render restored. I tried to make it integrate more with regular render but couldn't do it well, it still needs a 3D view to take the settings from, and can't run in a separate thread due to OpenGL. However, it is now rendering to an offscreen buffer which then gets displayed in the image window. This requires FBO's to be available, so a fallback creating a new window is still needed. Currently available from the Render menu in the top header.
2009-10-28 18:03:04 +00:00
/* GPUOffScreen */
struct GPUOffScreen {
GPUFrameBuffer *fb;
GPUTexture *color;
GPUTexture *depth;
};
Fix: OpenGL renders on graphics cards which do not support non-power-of-two textures were stretched and the wrong size.
2011-10-28 16:57:06 +00:00
GPUOffScreen *GPU_offscreen_create(int width, int height, char err_out[256])
OpenGL Render restored. I tried to make it integrate more with regular render but couldn't do it well, it still needs a 3D view to take the settings from, and can't run in a separate thread due to OpenGL. However, it is now rendering to an offscreen buffer which then gets displayed in the image window. This requires FBO's to be available, so a fallback creating a new window is still needed. Currently available from the Render menu in the top header.
2009-10-28 18:03:04 +00:00
{
GPUOffScreen *ofs;
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
ofs = MEM_callocN(sizeof(GPUOffScreen), "GPUOffScreen");
OpenGL Render restored. I tried to make it integrate more with regular render but couldn't do it well, it still needs a 3D view to take the settings from, and can't run in a separate thread due to OpenGL. However, it is now rendering to an offscreen buffer which then gets displayed in the image window. This requires FBO's to be available, so a fallback creating a new window is still needed. Currently available from the Render menu in the top header.
2009-10-28 18:03:04 +00:00
ofs->fb = GPU_framebuffer_create();
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (!ofs->fb) {
OpenGL Render restored. I tried to make it integrate more with regular render but couldn't do it well, it still needs a 3D view to take the settings from, and can't run in a separate thread due to OpenGL. However, it is now rendering to an offscreen buffer which then gets displayed in the image window. This requires FBO's to be available, so a fallback creating a new window is still needed. Currently available from the Render menu in the top header.
2009-10-28 18:03:04 +00:00
GPU_offscreen_free(ofs);
return NULL;
}
Fix: OpenGL renders on graphics cards which do not support non-power-of-two textures were stretched and the wrong size.
2011-10-28 16:57:06 +00:00
ofs->depth = GPU_texture_create_depth(width, height, err_out);
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (!ofs->depth) {
OpenGL Render restored. I tried to make it integrate more with regular render but couldn't do it well, it still needs a 3D view to take the settings from, and can't run in a separate thread due to OpenGL. However, it is now rendering to an offscreen buffer which then gets displayed in the image window. This requires FBO's to be available, so a fallback creating a new window is still needed. Currently available from the Render menu in the top header.
2009-10-28 18:03:04 +00:00
GPU_offscreen_free(ofs);
return NULL;
}
GPU framebuffer API: Allow binding a texture to a different texture attachment than the first. Also fix a number error in seperable gaussian blur shader.
2014-11-18 11:56:23 +01:00
if (!GPU_framebuffer_texture_attach(ofs->fb, ofs->depth, 0, err_out)) {
OpenGL Render restored. I tried to make it integrate more with regular render but couldn't do it well, it still needs a 3D view to take the settings from, and can't run in a separate thread due to OpenGL. However, it is now rendering to an offscreen buffer which then gets displayed in the image window. This requires FBO's to be available, so a fallback creating a new window is still needed. Currently available from the Render menu in the top header.
2009-10-28 18:03:04 +00:00
GPU_offscreen_free(ofs);
return NULL;
}
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
ofs->color = GPU_texture_create_2D(width, height, NULL, GPU_HDR_NONE, err_out);
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (!ofs->color) {
OpenGL Render restored. I tried to make it integrate more with regular render but couldn't do it well, it still needs a 3D view to take the settings from, and can't run in a separate thread due to OpenGL. However, it is now rendering to an offscreen buffer which then gets displayed in the image window. This requires FBO's to be available, so a fallback creating a new window is still needed. Currently available from the Render menu in the top header.
2009-10-28 18:03:04 +00:00
GPU_offscreen_free(ofs);
return NULL;
}
GPU framebuffer API: Allow binding a texture to a different texture attachment than the first. Also fix a number error in seperable gaussian blur shader.
2014-11-18 11:56:23 +01:00
if (!GPU_framebuffer_texture_attach(ofs->fb, ofs->color, 0, err_out)) {
OpenGL Render restored. I tried to make it integrate more with regular render but couldn't do it well, it still needs a 3D view to take the settings from, and can't run in a separate thread due to OpenGL. However, it is now rendering to an offscreen buffer which then gets displayed in the image window. This requires FBO's to be available, so a fallback creating a new window is still needed. Currently available from the Render menu in the top header.
2009-10-28 18:03:04 +00:00
GPU_offscreen_free(ofs);
return NULL;
}
Fix framebuffer completeness being broken after last framebuffer cleanup commits. Basically, we don't set a draw buffer until draw time comes. Also add explicit validation function to validate after all textures have been attached (could be done automatically at bind time too probably, but left out for now)
2014-11-25 16:16:50 +01:00
/* check validity at the very end! */
if (!GPU_framebuffer_check_valid(ofs->fb, err_out)) {
GPU_offscreen_free(ofs);
return NULL;
}
OpenGL Render restored. I tried to make it integrate more with regular render but couldn't do it well, it still needs a 3D view to take the settings from, and can't run in a separate thread due to OpenGL. However, it is now rendering to an offscreen buffer which then gets displayed in the image window. This requires FBO's to be available, so a fallback creating a new window is still needed. Currently available from the Render menu in the top header.
2009-10-28 18:03:04 +00:00
GPU_framebuffer_restore();
return ofs;
}
void GPU_offscreen_free(GPUOffScreen *ofs)
{
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (ofs->fb)
OpenGL Render restored. I tried to make it integrate more with regular render but couldn't do it well, it still needs a 3D view to take the settings from, and can't run in a separate thread due to OpenGL. However, it is now rendering to an offscreen buffer which then gets displayed in the image window. This requires FBO's to be available, so a fallback creating a new window is still needed. Currently available from the Render menu in the top header.
2009-10-28 18:03:04 +00:00
GPU_framebuffer_free(ofs->fb);
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (ofs->color)
OpenGL Render restored. I tried to make it integrate more with regular render but couldn't do it well, it still needs a 3D view to take the settings from, and can't run in a separate thread due to OpenGL. However, it is now rendering to an offscreen buffer which then gets displayed in the image window. This requires FBO's to be available, so a fallback creating a new window is still needed. Currently available from the Render menu in the top header.
2009-10-28 18:03:04 +00:00
GPU_texture_free(ofs->color);
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (ofs->depth)
OpenGL Render restored. I tried to make it integrate more with regular render but couldn't do it well, it still needs a 3D view to take the settings from, and can't run in a separate thread due to OpenGL. However, it is now rendering to an offscreen buffer which then gets displayed in the image window. This requires FBO's to be available, so a fallback creating a new window is still needed. Currently available from the Render menu in the top header.
2009-10-28 18:03:04 +00:00
GPU_texture_free(ofs->depth);
MEM_freeN(ofs);
}
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
void GPU_offscreen_bind(GPUOffScreen *ofs, bool save)
OpenGL Render restored. I tried to make it integrate more with regular render but couldn't do it well, it still needs a 3D view to take the settings from, and can't run in a separate thread due to OpenGL. However, it is now rendering to an offscreen buffer which then gets displayed in the image window. This requires FBO's to be available, so a fallback creating a new window is still needed. Currently available from the Render menu in the top header.
2009-10-28 18:03:04 +00:00
{
glDisable(GL_SCISSOR_TEST);
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
if (save)
Attempt to fix T43919, avoid using GPU_framebuffer_slots_bind just in case a system does not support drawing to many framebuffers.
2015-03-09 18:27:27 +01:00
GPU_texture_bind_as_framebuffer(ofs->color);
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
else {
GPU_framebuffer_bind_no_save(ofs->fb, 0);
}
OpenGL Render restored. I tried to make it integrate more with regular render but couldn't do it well, it still needs a 3D view to take the settings from, and can't run in a separate thread due to OpenGL. However, it is now rendering to an offscreen buffer which then gets displayed in the image window. This requires FBO's to be available, so a fallback creating a new window is still needed. Currently available from the Render menu in the top header.
2009-10-28 18:03:04 +00:00
}
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
void GPU_offscreen_unbind(GPUOffScreen *ofs, bool restore)
OpenGL Render restored. I tried to make it integrate more with regular render but couldn't do it well, it still needs a 3D view to take the settings from, and can't run in a separate thread due to OpenGL. However, it is now rendering to an offscreen buffer which then gets displayed in the image window. This requires FBO's to be available, so a fallback creating a new window is still needed. Currently available from the Render menu in the top header.
2009-10-28 18:03:04 +00:00
{
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
if (restore)
GPU_framebuffer_texture_unbind(ofs->fb, ofs->color);
OpenGL Render restored. I tried to make it integrate more with regular render but couldn't do it well, it still needs a 3D view to take the settings from, and can't run in a separate thread due to OpenGL. However, it is now rendering to an offscreen buffer which then gets displayed in the image window. This requires FBO's to be available, so a fallback creating a new window is still needed. Currently available from the Render menu in the top header.
2009-10-28 18:03:04 +00:00
GPU_framebuffer_restore();
glEnable(GL_SCISSOR_TEST);
}
Fix: OpenGL renders on graphics cards which do not support non-power-of-two textures were stretched and the wrong size.
2011-10-28 16:57:06 +00:00
void GPU_offscreen_read_pixels(GPUOffScreen *ofs, int type, void *pixels)
{
Alternate fix for offscreen render w/o npot Store the original texture size, since its needed for reading pixels and passing to glViewport.
2015-09-14 13:18:43 +10:00
glReadPixels(0, 0, ofs->color->w_orig, ofs->color->h_orig, GL_RGBA, type, pixels);
Fix: OpenGL renders on graphics cards which do not support non-power-of-two textures were stretched and the wrong size.
2011-10-28 16:57:06 +00:00
}
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
int GPU_offscreen_width(const GPUOffScreen *ofs)
Fix: multisample viewport drawing didn't work well with selection or particle brushes, due to issues with color coded drawing or slow/buggy reading from such a buffer on some systems. In case multisample is enabled now, it uses an offscreen buffer for such drawing, which is not multisampled and so should not cause issues. This does mean there is some extra GPU memory usage when multisample is enabled, and we could optimize triple buffer to work together here somehow to share buffers, but it's better than having selection not working.
2013-03-15 19:56:33 +00:00
{
Alternate fix for offscreen render w/o npot Store the original texture size, since its needed for reading pixels and passing to glViewport.
2015-09-14 13:18:43 +10:00
return ofs->color->w_orig;
Fix: multisample viewport drawing didn't work well with selection or particle brushes, due to issues with color coded drawing or slow/buggy reading from such a buffer on some systems. In case multisample is enabled now, it uses an offscreen buffer for such drawing, which is not multisampled and so should not cause issues. This does mean there is some extra GPU memory usage when multisample is enabled, and we could optimize triple buffer to work together here somehow to share buffers, but it's better than having selection not working.
2013-03-15 19:56:33 +00:00
}
Cleanup: bool & const Using bool when we're asking yes/no questions such as whether some GPU feature is supported. Consolidated these simple functions into gpu_extensions.c and grouped them in the header. Const-ified some args where the functions don't modify the pointed-to data.
2015-03-16 22:57:56 -04:00
int GPU_offscreen_height(const GPUOffScreen *ofs)
Fix: multisample viewport drawing didn't work well with selection or particle brushes, due to issues with color coded drawing or slow/buggy reading from such a buffer on some systems. In case multisample is enabled now, it uses an offscreen buffer for such drawing, which is not multisampled and so should not cause issues. This does mean there is some extra GPU memory usage when multisample is enabled, and we could optimize triple buffer to work together here somehow to share buffers, but it's better than having selection not working.
2013-03-15 19:56:33 +00:00
{
Alternate fix for offscreen render w/o npot Store the original texture size, since its needed for reading pixels and passing to glViewport.
2015-09-14 13:18:43 +10:00
return ofs->color->h_orig;
Fix: multisample viewport drawing didn't work well with selection or particle brushes, due to issues with color coded drawing or slow/buggy reading from such a buffer on some systems. In case multisample is enabled now, it uses an offscreen buffer for such drawing, which is not multisampled and so should not cause issues. This does mean there is some extra GPU memory usage when multisample is enabled, and we could optimize triple buffer to work together here somehow to share buffers, but it's better than having selection not working.
2013-03-15 19:56:33 +00:00
}
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
/* GPUShader */
struct GPUShader {
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
GLhandleARB object; /* handle for full shader */
GLhandleARB vertex; /* handle for vertex shader */
GLhandleARB fragment; /* handle for fragment shader */
GLhandleARB geometry; /* handle for geometry shader */
GLhandleARB lib; /* handle for libment shader */
int totattrib; /* total number of attributes */
int uniforms; /* required uniforms */
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
};
Fix T31546 fragment program gets created every frame That was really crappy indeed. Now we have a separate API for low level OpenGL programs, plus a nice interface for GPU, also removes some GL calls from main code as a plus :) The source for the programs is also moved to nice external .glsl files (not sure which extension convention GPU assemply uses)
2015-04-09 20:20:33 +02:00
struct GPUProgram {
GPUProgramType type;
GLuint prog;
};
static void shader_print_errors(const char *task, const char *log, const char **code, int totcode)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
Fix T41596 GLSL error on ATIs after clipping workaround commit. This was a little difficult to track down, basically it was a missing escape sequence that only manifested itself when GPU did not support bicubic filtering. Extra: * Fix memory leaks when an error occurs in shader compilation * Display full shader when a compilation error occurs. Makes it easier to diagnose if problem is caused by a syntax or compatibility error.
2014-08-29 16:23:37 +02:00
int i;
Shader errors: Count line from beginning of the whole shader source instead of each string sepatately since it helps with finding out the error line in most tested platforms
2015-03-23 15:19:54 +01:00
int line = 1;
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
fprintf(stderr, "GPUShader: %s error:\n", task);
Fix T41596 GLSL error on ATIs after clipping workaround commit. This was a little difficult to track down, basically it was a missing escape sequence that only manifested itself when GPU did not support bicubic filtering. Extra: * Fix memory leaks when an error occurs in shader compilation * Display full shader when a compilation error occurs. Makes it easier to diagnose if problem is caused by a syntax or compatibility error.
2014-08-29 16:23:37 +02:00
for (i = 0; i < totcode; i++) {
const char *c, *pos, *end = code[i] + strlen(code[i]);
GPU: code cleanup, no fundamental changes
2014-12-03 17:26:43 -02:00
if ((G.debug & G_DEBUG)) {
Fix T41596 GLSL error on ATIs after clipping workaround commit. This was a little difficult to track down, basically it was a missing escape sequence that only manifested itself when GPU did not support bicubic filtering. Extra: * Fix memory leaks when an error occurs in shader compilation * Display full shader when a compilation error occurs. Makes it easier to diagnose if problem is caused by a syntax or compatibility error.
2014-08-29 16:23:37 +02:00
fprintf(stderr, "===== shader string %d ====\n", i + 1);
c = code[i];
while ((c < end) && (pos = strchr(c, '\n'))) {
fprintf(stderr, "%2d ", line);
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
fwrite(c, (pos + 1) - c, 1, stderr);
c = pos + 1;
Fix T41596 GLSL error on ATIs after clipping workaround commit. This was a little difficult to track down, basically it was a missing escape sequence that only manifested itself when GPU did not support bicubic filtering. Extra: * Fix memory leaks when an error occurs in shader compilation * Display full shader when a compilation error occurs. Makes it easier to diagnose if problem is caused by a syntax or compatibility error.
2014-08-29 16:23:37 +02:00
line++;
}
fprintf(stderr, "%s", c);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
}
Fix T41596 GLSL error on ATIs after clipping workaround commit. This was a little difficult to track down, basically it was a missing escape sequence that only manifested itself when GPU did not support bicubic filtering. Extra: * Fix memory leaks when an error occurs in shader compilation * Display full shader when a compilation error occurs. Makes it easier to diagnose if problem is caused by a syntax or compatibility error.
2014-08-29 16:23:37 +02:00
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
fprintf(stderr, "%s\n", log);
}
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
static const char *gpu_shader_version(bool use_opensubdiv)
Another attempt for T40981, clipping border does not work with GLSL on ATIs. This is actually a test to see if this can be enabled on ATI cards. According to various sources, newer ATI cards supporting GLSL 3.0 support gl_ClippingDistance in shaders, which is the forward compatible way to do custom clipping. This fix will bind 6 additional varying variables on ATIs, which may lead to some shaders not compiling due to limiting out of those variables, or to performance degradation. Also I do not have an ATI handy to test. Having those in mind, this commit may well be reverted later. Clipping planes are usually 4 (6 is for cube clipping), but making shaders depend on viewport state is really bad, and would lead to recompilation, so I took the worst case here to avoid that. Hopefully driver does some optimization there.
2014-07-11 19:17:29 +03:00
{
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
#ifdef WITH_OPENSUBDIV
if (use_opensubdiv) {
OpenSubdiv: Corrections to GLSL version Was own mistake in the last minute fixes, now matcaps should work just fine.
2015-08-03 15:27:06 +02:00
return "#version 130\n";
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
}
Cleanup: warning
2015-07-21 09:54:23 +10:00
#else
UNUSED_VARS(use_opensubdiv);
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
#endif
Another attempt for T40981, clipping border does not work with GLSL on ATIs. This is actually a test to see if this can be enabled on ATI cards. According to various sources, newer ATI cards supporting GLSL 3.0 support gl_ClippingDistance in shaders, which is the forward compatible way to do custom clipping. This fix will bind 6 additional varying variables on ATIs, which may lead to some shaders not compiling due to limiting out of those variables, or to performance degradation. Also I do not have an ATI handy to test. Having those in mind, this commit may well be reverted later. Clipping planes are usually 4 (6 is for cube clipping), but making shaders depend on viewport state is really bad, and would lead to recompilation, so I took the worst case here to avoid that. Hopefully driver does some optimization there.
2014-07-11 19:17:29 +03:00
/* turn on glsl 1.30 for bicubic bump mapping and ATI clipping support */
if (GLEW_VERSION_3_0 &&
(GPU_bicubic_bump_support() || GPU_type_matches(GPU_DEVICE_ATI, GPU_OS_ANY, GPU_DRIVER_ANY)))
{
return "#version 130\n";
}
return "";
}
Fix T45491 not all GPUs support those extensions, use check for opensubdiv to enable.
2015-08-20 16:05:24 +03:00
static void gpu_shader_standard_extensions(char defines[MAX_EXT_DEFINE_LENGTH], bool use_opensubdiv)
Fix #34788, #34744: GLSL error, #version line needs to be at the top of the shader and this wasn't the case anymore after recent changes.
2013-04-02 16:37:31 +00:00
{
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
#ifdef WITH_OPENSUBDIV
Fix T45491 not all GPUs support those extensions, use check for opensubdiv to enable.
2015-08-20 16:05:24 +03:00
if (use_opensubdiv) {
strcat(defines, "#extension GL_ARB_texture_query_lod: enable\n"
"#extension GL_ARB_gpu_shader5 : enable\n"
"#extension GL_ARB_explicit_attrib_location : require\n");
}
Fix T45860, missing extension in GLSL after recent own fix.
2015-08-23 18:20:08 +03:00
else if (GPU_bicubic_bump_support())
strcat(defines, "#extension GL_ARB_texture_query_lod: enable\n");
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
#else
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
/* need this extension for high quality bump mapping */
Another attempt for T40981, clipping border does not work with GLSL on ATIs. This is actually a test to see if this can be enabled on ATI cards. According to various sources, newer ATI cards supporting GLSL 3.0 support gl_ClippingDistance in shaders, which is the forward compatible way to do custom clipping. This fix will bind 6 additional varying variables on ATIs, which may lead to some shaders not compiling due to limiting out of those variables, or to performance degradation. Also I do not have an ATI handy to test. Having those in mind, this commit may well be reverted later. Clipping planes are usually 4 (6 is for cube clipping), but making shaders depend on viewport state is really bad, and would lead to recompilation, so I took the worst case here to avoid that. Hopefully driver does some optimization there.
2014-07-11 19:17:29 +03:00
if (GPU_bicubic_bump_support())
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
strcat(defines, "#extension GL_ARB_texture_query_lod: enable\n");
Fix T45491 not all GPUs support those extensions, use check for opensubdiv to enable.
2015-08-20 16:05:24 +03:00
(void) use_opensubdiv;
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
#endif
Fix #34788, #34744: GLSL error, #version line needs to be at the top of the shader and this wasn't the case anymore after recent changes.
2013-04-02 16:37:31 +00:00
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
if (GPU_geometry_shader_support())
strcat(defines, "#extension GL_EXT_geometry_shader4: enable\n");
if (GPU_instanced_drawing_support()) {
strcat(defines, "#extension GL_EXT_gpu_shader4: enable\n");
Fix high quality depth of field on the Mac. Quite a few things wrong here: * Mac did not support EXT_draw_instanced, only ARB_draw_instanced * Draw instanced did not work unless data came from vertex buffer, which is second time we see weird things with vertex arrays in mac * There were a few stupid mistakes by me as well, such as binding to uniform locations for the wrong shaders (it's a wonder it ever worked :p)
2015-03-20 15:26:13 +01:00
strcat(defines, "#extension GL_ARB_draw_instanced: enable\n");
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
}
Fix #34788, #34744: GLSL error, #version line needs to be at the top of the shader and this wasn't the case anymore after recent changes.
2013-04-02 16:37:31 +00:00
}
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
static void gpu_shader_standard_defines(bool use_opensubdiv,
char defines[MAX_DEFINE_LENGTH])
Fix #34492: clipping border not working with GLSL/matcap and Nouveau drivers.
2013-03-13 18:00:13 +00:00
{
/* some useful defines to detect GPU type */
Another attempt for T40981, clipping border does not work with GLSL on ATIs. This is actually a test to see if this can be enabled on ATI cards. According to various sources, newer ATI cards supporting GLSL 3.0 support gl_ClippingDistance in shaders, which is the forward compatible way to do custom clipping. This fix will bind 6 additional varying variables on ATIs, which may lead to some shaders not compiling due to limiting out of those variables, or to performance degradation. Also I do not have an ATI handy to test. Having those in mind, this commit may well be reverted later. Clipping planes are usually 4 (6 is for cube clipping), but making shaders depend on viewport state is really bad, and would lead to recompilation, so I took the worst case here to avoid that. Hopefully driver does some optimization there.
2014-07-11 19:17:29 +03:00
if (GPU_type_matches(GPU_DEVICE_ATI, GPU_OS_ANY, GPU_DRIVER_ANY)) {
strcat(defines, "#define GPU_ATI\n");
if (GLEW_VERSION_3_0)
Fix T41596 GLSL error on ATIs after clipping workaround commit. This was a little difficult to track down, basically it was a missing escape sequence that only manifested itself when GPU did not support bicubic filtering. Extra: * Fix memory leaks when an error occurs in shader compilation * Display full shader when a compilation error occurs. Makes it easier to diagnose if problem is caused by a syntax or compatibility error.
2014-08-29 16:23:37 +02:00
strcat(defines, "#define CLIP_WORKAROUND\n");
Another attempt for T40981, clipping border does not work with GLSL on ATIs. This is actually a test to see if this can be enabled on ATI cards. According to various sources, newer ATI cards supporting GLSL 3.0 support gl_ClippingDistance in shaders, which is the forward compatible way to do custom clipping. This fix will bind 6 additional varying variables on ATIs, which may lead to some shaders not compiling due to limiting out of those variables, or to performance degradation. Also I do not have an ATI handy to test. Having those in mind, this commit may well be reverted later. Clipping planes are usually 4 (6 is for cube clipping), but making shaders depend on viewport state is really bad, and would lead to recompilation, so I took the worst case here to avoid that. Hopefully driver does some optimization there.
2014-07-11 19:17:29 +03:00
}
code cleanup: also fix crash in GPU_state_print(). and confine to debug mode builds.
2013-06-13 11:59:28 +00:00
else if (GPU_type_matches(GPU_DEVICE_NVIDIA, GPU_OS_ANY, GPU_DRIVER_ANY))
Another attempt for T40981, clipping border does not work with GLSL on ATIs. This is actually a test to see if this can be enabled on ATI cards. According to various sources, newer ATI cards supporting GLSL 3.0 support gl_ClippingDistance in shaders, which is the forward compatible way to do custom clipping. This fix will bind 6 additional varying variables on ATIs, which may lead to some shaders not compiling due to limiting out of those variables, or to performance degradation. Also I do not have an ATI handy to test. Having those in mind, this commit may well be reverted later. Clipping planes are usually 4 (6 is for cube clipping), but making shaders depend on viewport state is really bad, and would lead to recompilation, so I took the worst case here to avoid that. Hopefully driver does some optimization there.
2014-07-11 19:17:29 +03:00
strcat(defines, "#define GPU_NVIDIA\n");
code cleanup: also fix crash in GPU_state_print(). and confine to debug mode builds.
2013-06-13 11:59:28 +00:00
else if (GPU_type_matches(GPU_DEVICE_INTEL, GPU_OS_ANY, GPU_DRIVER_ANY))
Another attempt for T40981, clipping border does not work with GLSL on ATIs. This is actually a test to see if this can be enabled on ATI cards. According to various sources, newer ATI cards supporting GLSL 3.0 support gl_ClippingDistance in shaders, which is the forward compatible way to do custom clipping. This fix will bind 6 additional varying variables on ATIs, which may lead to some shaders not compiling due to limiting out of those variables, or to performance degradation. Also I do not have an ATI handy to test. Having those in mind, this commit may well be reverted later. Clipping planes are usually 4 (6 is for cube clipping), but making shaders depend on viewport state is really bad, and would lead to recompilation, so I took the worst case here to avoid that. Hopefully driver does some optimization there.
2014-07-11 19:17:29 +03:00
strcat(defines, "#define GPU_INTEL\n");
if (GPU_bicubic_bump_support())
strcat(defines, "#define BUMP_BICUBIC\n");
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
#ifdef WITH_OPENSUBDIV
/* TODO(sergey): Check whether we actually compiling shader for
* the OpenSubdiv mesh.
*/
if (use_opensubdiv) {
strcat(defines, "#define USE_OPENSUBDIV\n");
/* TODO(sergey): not strictly speaking a define, but this is
* a global typedef which we don't have better place to define
* in yet.
*/
strcat(defines, "struct VertexData {\n"
" vec4 position;\n"
" vec3 normal;\n"
" vec2 uv;"
"};\n");
}
#else
UNUSED_VARS(use_opensubdiv);
#endif
Another attempt for T40981, clipping border does not work with GLSL on ATIs. This is actually a test to see if this can be enabled on ATI cards. According to various sources, newer ATI cards supporting GLSL 3.0 support gl_ClippingDistance in shaders, which is the forward compatible way to do custom clipping. This fix will bind 6 additional varying variables on ATIs, which may lead to some shaders not compiling due to limiting out of those variables, or to performance degradation. Also I do not have an ATI handy to test. Having those in mind, this commit may well be reverted later. Clipping planes are usually 4 (6 is for cube clipping), but making shaders depend on viewport state is really bad, and would lead to recompilation, so I took the worst case here to avoid that. Hopefully driver does some optimization there.
2014-07-11 19:17:29 +03:00
return;
Fix #34492: clipping border not working with GLSL/matcap and Nouveau drivers.
2013-03-13 18:00:13 +00:00
}
Fix T31546 fragment program gets created every frame That was really crappy indeed. Now we have a separate API for low level OpenGL programs, plus a nice interface for GPU, also removes some GL calls from main code as a plus :) The source for the programs is also moved to nice external .glsl files (not sure which extension convention GPU assemply uses)
2015-04-09 20:20:33 +02:00
void GPU_program_bind(GPUProgram *program)
{
glEnable(program->type);
glBindProgramARB(program->type, program->prog);
}
void GPU_program_unbind(GPUProgram *program)
{
glDisable(program->type);
glBindProgramARB(program->type, 0);
}
GPUProgram *GPU_program_shader_create(GPUProgramType type, const char *code)
{
GPUProgram *program;
GLint error_pos, is_native;
if (!(GLEW_ARB_fragment_program && type == GPU_PROGRAM_TYPE_FRAGMENT))
return NULL;
program = MEM_callocN(sizeof(GPUProgram), "GPUProgram");
switch (type) {
case GPU_PROGRAM_TYPE_FRAGMENT:
program->type = GL_FRAGMENT_PROGRAM_ARB;
break;
}
/* create the object and set its code string */
glGenProgramsARB(1, &program->prog);
glBindProgramARB(program->type, program->prog);
glProgramStringARB(program->type, GL_PROGRAM_FORMAT_ASCII_ARB, (GLsizei)strlen(code), code);
glGetIntegerv(GL_PROGRAM_ERROR_POSITION_ARB, &error_pos);
glGetProgramivARB(GL_FRAGMENT_PROGRAM_ARB, GL_PROGRAM_UNDER_NATIVE_LIMITS_ARB, &is_native);
if ((error_pos == -1) && (is_native == 1)) {
return program;
}
else {
/* glGetError is set before that, clear it */
while (glGetError() != GL_NO_ERROR)
;
shader_print_errors("compile", (const char *)glGetString(GL_PROGRAM_ERROR_STRING_ARB), &code, 1);
MEM_freeN(program);
}
return NULL;
}
void GPU_program_free(GPUProgram *program)
{
glDeleteProgramsARB(1, &program->prog);
MEM_freeN(program);
}
void GPU_program_parameter_4f(GPUProgram *program, unsigned int location, float x, float y, float z, float w)
{
glProgramLocalParameter4fARB(program->type, location, x, y, z, w);
}
Fix T46128: High Quality DoF broken The issue was caused by some special tricks needed to compile OpenSubdiv shader which was using stupid check whether geometry shader is used or not. Now made it more explicit call whether special OpenSubdiv trickery is needed or not. Its not ideal solution, but it's not really easy to do a proper solution for this, because while we can do half of the work with if-defs in the shader code but we'll still need to somewhat define layout of the input blocks which isn't really doable with current shader version we're using.
2015-09-16 22:38:46 +05:00
GPUShader *GPU_shader_create(const char *vertexcode,
const char *fragcode,
const char *geocode,
const char *libcode,
const char *defines,
int input,
int output,
int number)
{
return GPU_shader_create_ex(vertexcode,
fragcode,
geocode,
libcode,
defines,
input,
output,
number,
GPU_SHADER_FLAGS_NONE);
}
GPUShader *GPU_shader_create_ex(const char *vertexcode,
const char *fragcode,
const char *geocode,
const char *libcode,
const char *defines,
int input,
int output,
int number,
const int flags)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
OpenSubdiv: Corrections to GLSL version Was own mistake in the last minute fixes, now matcaps should work just fine.
2015-08-03 15:27:06 +02:00
#ifdef WITH_OPENSUBDIV
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
/* TODO(sergey): used to add #version 150 to the geometry shader.
Cleanup: spelling
2015-09-14 02:21:15 +10:00
* Could safely be renamed to "use_geometry_code" since it's very
* likely any of geometry code will want to use GLSL 1.5.
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
*/
Fix T46128: High Quality DoF broken The issue was caused by some special tricks needed to compile OpenSubdiv shader which was using stupid check whether geometry shader is used or not. Now made it more explicit call whether special OpenSubdiv trickery is needed or not. Its not ideal solution, but it's not really easy to do a proper solution for this, because while we can do half of the work with if-defs in the shader code but we'll still need to somewhat define layout of the input blocks which isn't really doable with current shader version we're using.
2015-09-16 22:38:46 +05:00
bool use_opensubdiv = (flags & GPU_SHADER_FLAGS_SPECIAL_OPENSUBDIV) != 0;
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
#else
bool use_opensubdiv = false;
#endif
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
GLint status;
GLcharARB log[5000];
GLsizei length = 0;
GPUShader *shader;
Another attempt for T40981, clipping border does not work with GLSL on ATIs. This is actually a test to see if this can be enabled on ATI cards. According to various sources, newer ATI cards supporting GLSL 3.0 support gl_ClippingDistance in shaders, which is the forward compatible way to do custom clipping. This fix will bind 6 additional varying variables on ATIs, which may lead to some shaders not compiling due to limiting out of those variables, or to performance degradation. Also I do not have an ATI handy to test. Having those in mind, this commit may well be reverted later. Clipping planes are usually 4 (6 is for cube clipping), but making shaders depend on viewport state is really bad, and would lead to recompilation, so I took the worst case here to avoid that. Hopefully driver does some optimization there.
2014-07-11 19:17:29 +03:00
char standard_defines[MAX_DEFINE_LENGTH] = "";
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
char standard_extensions[MAX_EXT_DEFINE_LENGTH] = "";
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
Cleanup: warning, style
2015-09-21 06:59:50 +10:00
#ifndef WITH_OPENSUBDIV
UNUSED_VARS(flags);
#endif
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
if (!GLEW_ARB_vertex_shader || !GLEW_ARB_fragment_shader || (geocode && !GPU_geometry_shader_support()))
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
return NULL;
shader = MEM_callocN(sizeof(GPUShader), "GPUShader");
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (vertexcode)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
shader->vertex = glCreateShaderObjectARB(GL_VERTEX_SHADER_ARB);
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (fragcode)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
shader->fragment = glCreateShaderObjectARB(GL_FRAGMENT_SHADER_ARB);
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
if (geocode)
shader->geometry = glCreateShaderObjectARB(GL_GEOMETRY_SHADER_EXT);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
shader->object = glCreateProgramObjectARB();
if (!shader->object ||
style cleanup: brace placement/newlines
2012-05-24 16:35:45 +00:00
(vertexcode && !shader->vertex) ||
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
(fragcode && !shader->fragment) ||
(geocode && !shader->geometry))
style cleanup: brace placement/newlines
2012-05-24 16:35:45 +00:00
{
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
fprintf(stderr, "GPUShader, object creation failed.\n");
GPU_shader_free(shader);
return NULL;
}
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
gpu_shader_standard_defines(use_opensubdiv, standard_defines);
Fix T45491 not all GPUs support those extensions, use check for opensubdiv to enable.
2015-08-20 16:05:24 +03:00
gpu_shader_standard_extensions(standard_extensions, use_opensubdiv);
Another attempt for T40981, clipping border does not work with GLSL on ATIs. This is actually a test to see if this can be enabled on ATI cards. According to various sources, newer ATI cards supporting GLSL 3.0 support gl_ClippingDistance in shaders, which is the forward compatible way to do custom clipping. This fix will bind 6 additional varying variables on ATIs, which may lead to some shaders not compiling due to limiting out of those variables, or to performance degradation. Also I do not have an ATI handy to test. Having those in mind, this commit may well be reverted later. Clipping planes are usually 4 (6 is for cube clipping), but making shaders depend on viewport state is really bad, and would lead to recompilation, so I took the worst case here to avoid that. Hopefully driver does some optimization there.
2014-07-11 19:17:29 +03:00
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (vertexcode) {
Another attempt for T40981, clipping border does not work with GLSL on ATIs. This is actually a test to see if this can be enabled on ATI cards. According to various sources, newer ATI cards supporting GLSL 3.0 support gl_ClippingDistance in shaders, which is the forward compatible way to do custom clipping. This fix will bind 6 additional varying variables on ATIs, which may lead to some shaders not compiling due to limiting out of those variables, or to performance degradation. Also I do not have an ATI handy to test. Having those in mind, this commit may well be reverted later. Clipping planes are usually 4 (6 is for cube clipping), but making shaders depend on viewport state is really bad, and would lead to recompilation, so I took the worst case here to avoid that. Hopefully driver does some optimization there.
2014-07-11 19:17:29 +03:00
const char *source[5];
/* custom limit, may be too small, beware */
OpenGL: implemenation of fixed function lighting as per pixel GLSL shaders. The code is still unused, but the intention is to use this to solve the double sided lighting problem on NVidia, and to make the materials work on OpenGL ES 2.0 eventually. The code works and matches the fixed function lighting pretty much exactly, but still needs optimizations. The actual integration in object draw will be committed later when more fixing & testing, there's lots of different combinations and unclear OpenGL state here.
2013-02-26 00:49:42 +00:00
int num_source = 0;
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
source[num_source++] = gpu_shader_version(use_opensubdiv);
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
source[num_source++] = standard_extensions;
Another attempt for T40981, clipping border does not work with GLSL on ATIs. This is actually a test to see if this can be enabled on ATI cards. According to various sources, newer ATI cards supporting GLSL 3.0 support gl_ClippingDistance in shaders, which is the forward compatible way to do custom clipping. This fix will bind 6 additional varying variables on ATIs, which may lead to some shaders not compiling due to limiting out of those variables, or to performance degradation. Also I do not have an ATI handy to test. Having those in mind, this commit may well be reverted later. Clipping planes are usually 4 (6 is for cube clipping), but making shaders depend on viewport state is really bad, and would lead to recompilation, so I took the worst case here to avoid that. Hopefully driver does some optimization there.
2014-07-11 19:17:29 +03:00
source[num_source++] = standard_defines;
Fix #34492: clipping border not working with GLSL/matcap and Nouveau drivers.
2013-03-13 18:00:13 +00:00
OpenGL: implemenation of fixed function lighting as per pixel GLSL shaders. The code is still unused, but the intention is to use this to solve the double sided lighting problem on NVidia, and to make the materials work on OpenGL ES 2.0 eventually. The code works and matches the fixed function lighting pretty much exactly, but still needs optimizations. The actual integration in object draw will be committed later when more fixing & testing, there's lots of different combinations and unclear OpenGL state here.
2013-02-26 00:49:42 +00:00
if (defines) source[num_source++] = defines;
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
source[num_source++] = vertexcode;
OpenGL: implemenation of fixed function lighting as per pixel GLSL shaders. The code is still unused, but the intention is to use this to solve the double sided lighting problem on NVidia, and to make the materials work on OpenGL ES 2.0 eventually. The code works and matches the fixed function lighting pretty much exactly, but still needs optimizations. The actual integration in object draw will be committed later when more fixing & testing, there's lots of different combinations and unclear OpenGL state here.
2013-02-26 00:49:42 +00:00
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
glAttachObjectARB(shader->object, shader->vertex);
OpenGL: implemenation of fixed function lighting as per pixel GLSL shaders. The code is still unused, but the intention is to use this to solve the double sided lighting problem on NVidia, and to make the materials work on OpenGL ES 2.0 eventually. The code works and matches the fixed function lighting pretty much exactly, but still needs optimizations. The actual integration in object draw will be committed later when more fixing & testing, there's lots of different combinations and unclear OpenGL state here.
2013-02-26 00:49:42 +00:00
glShaderSourceARB(shader->vertex, num_source, source, NULL);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
glCompileShaderARB(shader->vertex);
glGetObjectParameterivARB(shader->vertex, GL_OBJECT_COMPILE_STATUS_ARB, &status);
if (!status) {
glGetInfoLogARB(shader->vertex, sizeof(log), &length, log);
Fix T41596 GLSL error on ATIs after clipping workaround commit. This was a little difficult to track down, basically it was a missing escape sequence that only manifested itself when GPU did not support bicubic filtering. Extra: * Fix memory leaks when an error occurs in shader compilation * Display full shader when a compilation error occurs. Makes it easier to diagnose if problem is caused by a syntax or compatibility error.
2014-08-29 16:23:37 +02:00
shader_print_errors("compile", log, source, num_source);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
GPU_shader_free(shader);
return NULL;
}
}
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (fragcode) {
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
const char *source[7];
OpenGL: implemenation of fixed function lighting as per pixel GLSL shaders. The code is still unused, but the intention is to use this to solve the double sided lighting problem on NVidia, and to make the materials work on OpenGL ES 2.0 eventually. The code works and matches the fixed function lighting pretty much exactly, but still needs optimizations. The actual integration in object draw will be committed later when more fixing & testing, there's lots of different combinations and unclear OpenGL state here.
2013-02-26 00:49:42 +00:00
int num_source = 0;
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
source[num_source++] = gpu_shader_version(use_opensubdiv);
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
source[num_source++] = standard_extensions;
Another attempt for T40981, clipping border does not work with GLSL on ATIs. This is actually a test to see if this can be enabled on ATI cards. According to various sources, newer ATI cards supporting GLSL 3.0 support gl_ClippingDistance in shaders, which is the forward compatible way to do custom clipping. This fix will bind 6 additional varying variables on ATIs, which may lead to some shaders not compiling due to limiting out of those variables, or to performance degradation. Also I do not have an ATI handy to test. Having those in mind, this commit may well be reverted later. Clipping planes are usually 4 (6 is for cube clipping), but making shaders depend on viewport state is really bad, and would lead to recompilation, so I took the worst case here to avoid that. Hopefully driver does some optimization there.
2014-07-11 19:17:29 +03:00
source[num_source++] = standard_defines;
Fix #34492: clipping border not working with GLSL/matcap and Nouveau drivers.
2013-03-13 18:00:13 +00:00
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
#ifdef WITH_OPENSUBDIV
/* TODO(sergey): Move to fragment shader source code generation. */
if (use_opensubdiv) {
source[num_source++] =
"#ifdef USE_OPENSUBDIV\n"
"in block {\n"
" VertexData v;\n"
"} inpt;\n"
"#endif\n";
}
#endif
OpenGL: implemenation of fixed function lighting as per pixel GLSL shaders. The code is still unused, but the intention is to use this to solve the double sided lighting problem on NVidia, and to make the materials work on OpenGL ES 2.0 eventually. The code works and matches the fixed function lighting pretty much exactly, but still needs optimizations. The actual integration in object draw will be committed later when more fixing & testing, there's lots of different combinations and unclear OpenGL state here.
2013-02-26 00:49:42 +00:00
if (defines) source[num_source++] = defines;
if (libcode) source[num_source++] = libcode;
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
source[num_source++] = fragcode;
Fix again for the NVidia driver bug. This time I'm just giving up using linking together the precompiled library shader code and material code and recompiling it all again for each material even if it gives a performance hit, since the previous workaround only worked on some driver versions still.
2008-09-19 16:09:26 +00:00
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
glAttachObjectARB(shader->object, shader->fragment);
OpenGL: implemenation of fixed function lighting as per pixel GLSL shaders. The code is still unused, but the intention is to use this to solve the double sided lighting problem on NVidia, and to make the materials work on OpenGL ES 2.0 eventually. The code works and matches the fixed function lighting pretty much exactly, but still needs optimizations. The actual integration in object draw will be committed later when more fixing & testing, there's lots of different combinations and unclear OpenGL state here.
2013-02-26 00:49:42 +00:00
glShaderSourceARB(shader->fragment, num_source, source, NULL);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
glCompileShaderARB(shader->fragment);
glGetObjectParameterivARB(shader->fragment, GL_OBJECT_COMPILE_STATUS_ARB, &status);
if (!status) {
glGetInfoLogARB(shader->fragment, sizeof(log), &length, log);
Fix T41596 GLSL error on ATIs after clipping workaround commit. This was a little difficult to track down, basically it was a missing escape sequence that only manifested itself when GPU did not support bicubic filtering. Extra: * Fix memory leaks when an error occurs in shader compilation * Display full shader when a compilation error occurs. Makes it easier to diagnose if problem is caused by a syntax or compatibility error.
2014-08-29 16:23:37 +02:00
shader_print_errors("compile", log, source, num_source);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
GPU_shader_free(shader);
return NULL;
}
}
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
if (geocode) {
const char *source[6];
int num_source = 0;
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
source[num_source++] = gpu_shader_version(use_opensubdiv);
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
source[num_source++] = standard_extensions;
source[num_source++] = standard_defines;
if (defines) source[num_source++] = defines;
source[num_source++] = geocode;
glAttachObjectARB(shader->object, shader->geometry);
glShaderSourceARB(shader->geometry, num_source, source, NULL);
glCompileShaderARB(shader->geometry);
glGetObjectParameterivARB(shader->geometry, GL_OBJECT_COMPILE_STATUS_ARB, &status);
if (!status) {
glGetInfoLogARB(shader->geometry, sizeof(log), &length, log);
shader_print_errors("compile", log, source, num_source);
GPU_shader_free(shader);
return NULL;
}
Fix high quality depth of field on the Mac. Quite a few things wrong here: * Mac did not support EXT_draw_instanced, only ARB_draw_instanced * Draw instanced did not work unless data came from vertex buffer, which is second time we see weird things with vertex arrays in mac * There were a few stupid mistakes by me as well, such as binding to uniform locations for the wrong shaders (it's a wonder it ever worked :p)
2015-03-20 15:26:13 +01:00
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
if (!use_opensubdiv) {
GPU_shader_geometry_stage_primitive_io(shader, input, output, number);
}
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
}
style cleanup: comment blocks
2012-03-09 18:28:30 +00:00
#if 0
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (lib && lib->lib)
style cleanup: comment blocks
2012-03-09 18:28:30 +00:00
glAttachObjectARB(shader->object, lib->lib);
#endif
Workaround for an Nvidia driver bug on 32 bit linux (maybe windows too). I reversed the order of attaching shader objects for linking to solve an issue in the driver before, but now it appears it needs to be the other way around again? I don't know if these are the same cards that now want it different again due to changes in the glsl code, but I found another workaround for the first bug in a forum post (leaving out parameter names in the declarations), so with some luck both cases work? http://developer.nvidia.com/forums/index.php?showtopic=596
2008-09-15 18:15:11 +00:00
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
#ifdef WITH_OPENSUBDIV
if (use_opensubdiv) {
glBindAttribLocation(shader->object, 0, "position");
glBindAttribLocation(shader->object, 1, "normal");
GPU_shader_geometry_stage_primitive_io(shader,
GL_LINES_ADJACENCY_EXT,
GL_TRIANGLE_STRIP,
4);
}
#endif
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
glLinkProgramARB(shader->object);
glGetObjectParameterivARB(shader->object, GL_OBJECT_LINK_STATUS_ARB, &status);
if (!status) {
glGetInfoLogARB(shader->object, sizeof(log), &length, log);
Fix T41596 GLSL error on ATIs after clipping workaround commit. This was a little difficult to track down, basically it was a missing escape sequence that only manifested itself when GPU did not support bicubic filtering. Extra: * Fix memory leaks when an error occurs in shader compilation * Display full shader when a compilation error occurs. Makes it easier to diagnose if problem is caused by a syntax or compatibility error.
2014-08-29 16:23:37 +02:00
if (fragcode) shader_print_errors("linking", log, &fragcode, 1);
else if (vertexcode) shader_print_errors("linking", log, &vertexcode, 1);
else if (libcode) shader_print_errors("linking", log, &libcode, 1);
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
else if (geocode) shader_print_errors("linking", log, &geocode, 1);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
GPU_shader_free(shader);
return NULL;
}
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
#ifdef WITH_OPENSUBDIV
/* TODO(sergey): Find a better place for this. */
Fix T45691 crash on material view Call glProgramUniform1i only when OpenSubdiv is requested and only when the GPU supports it.
2015-08-06 01:09:07 +02:00
if (use_opensubdiv && GLEW_VERSION_4_1) {
OpenSubdiv: Commit of OpenSubdiv integration into Blender This commit contains all the remained parts needed for initial integration of OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU backends which works in the following way: - When SubSurf modifier is the last in the modifiers stack then GPU pipeline of OpenSubdiv is used, making viewport performance as fast as possible. This also requires graphscard with GLSL 1.5 support. If this requirement is not met, then no GPU pipeline is used at all. - If SubSurf is not a last modifier or if DerivesMesh is being evaluated for rendering then CPU limit evaluation API from OpenSubdiv is used. This only replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG structures exactly the same as they used to be for ages now. This integration is fully covered with ifdef and not enabled by default because there are several TODOs to be solved first: - Face varying data interpolation is not really cleanly implemented for GPU in OpenSubdiv 3.0. It is also not implemented for limit evaluation API. This basically means we'll have really hard time supporting UVs. - Limit evaluation only works with adaptivly subdivided meshes so far, which basically means all the points of CCG are pushed to the limit. This gives different result from old code. - There are some serious optimizations possible on the topology refiner creation, which would speed up initial OpenSubdiv mesh creation. - There are some hardcoded asumptions in the GPU and DerivedMesh areas which could be generalized. That's something where Antony and Campbell can help, making it so the code is structured in a way which is reusable by all planned viewport projects. - There are also some workarounds in the dependency graph to make sure OpenGL buffers are only freed from the main thread. Those who'll be wanting to make experiments with this code should grab dev branch (NOT master) from https://github.com/Nazg-Gul/OpenSubdiv/tree/dev There are some patches applied in there which we're working on on getting into upstream.
2015-07-20 16:08:06 +02:00
glProgramUniform1i(shader->object,
glGetUniformLocation(shader->object, "FVarDataBuffer"),
31); /* GL_TEXTURE31 */
}
#endif
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
return shader;
}
Fix again for the NVidia driver bug. This time I'm just giving up using linking together the precompiled library shader code and material code and recompiling it all again for each material even if it gives a performance hit, since the previous workaround only worked on some driver versions still.
2008-09-19 16:09:26 +00:00
#if 0
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
GPUShader *GPU_shader_create_lib(const char *code)
{
GLint status;
GLcharARB log[5000];
GLsizei length = 0;
GPUShader *shader;
if (!GLEW_ARB_vertex_shader || !GLEW_ARB_fragment_shader)
return NULL;
shader = MEM_callocN(sizeof(GPUShader), "GPUShader");
shader->lib = glCreateShaderObjectARB(GL_FRAGMENT_SHADER_ARB);
if (!shader->lib) {
fprintf(stderr, "GPUShader, object creation failed.\n");
GPU_shader_free(shader);
return NULL;
}
glShaderSourceARB(shader->lib, 1, (const char**)&code, NULL);
glCompileShaderARB(shader->lib);
glGetObjectParameterivARB(shader->lib, GL_OBJECT_COMPILE_STATUS_ARB, &status);
if (!status) {
glGetInfoLogARB(shader->lib, sizeof(log), &length, log);
shader_print_errors("compile", log, code);
GPU_shader_free(shader);
return NULL;
}
return shader;
}
Fix again for the NVidia driver bug. This time I'm just giving up using linking together the precompiled library shader code and material code and recompiling it all again for each material even if it gives a performance hit, since the previous workaround only worked on some driver versions still.
2008-09-19 16:09:26 +00:00
#endif
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
void GPU_shader_bind(GPUShader *shader)
{
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_ASSERT_NO_GL_ERRORS("Pre Shader Bind");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
glUseProgramObjectARB(shader->object);
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_ASSERT_NO_GL_ERRORS("Post Shader Bind");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
code cleanup: quiet struct gcc warnings, also use more conventional names for bmesh dissolve.
2013-03-12 08:50:02 +00:00
void GPU_shader_unbind(void)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_ASSERT_NO_GL_ERRORS("Pre Shader Unbind");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
glUseProgramObjectARB(0);
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_ASSERT_NO_GL_ERRORS("Post Shader Unbind");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
void GPU_shader_free(GPUShader *shader)
{
if (shader->lib)
glDeleteObjectARB(shader->lib);
if (shader->vertex)
glDeleteObjectARB(shader->vertex);
if (shader->fragment)
glDeleteObjectARB(shader->fragment);
if (shader->object)
glDeleteObjectARB(shader->object);
MEM_freeN(shader);
}
Enabled GCC -Wwrite-strings warning for CMake and replaced many 'char's for 'const char's,. Only one functional change where Transform orientations passed "" to BIF_createTransformOrientation() which could then have the value written into.
2010-12-03 17:05:21 +00:00
int GPU_shader_get_uniform(GPUShader *shader, const char *name)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
return glGetUniformLocationARB(shader->object, name);
}
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
void GPU_shader_uniform_vector(GPUShader *UNUSED(shader), int location, int length, int arraysize, const float *value)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
Attempt to fix T45718, crash when enabling GLSL mode with UV layers. I can't reproduce issue here but crash is most likely caused by passing a NULL pointer to glUniform2fv. This is caused by OpenSubdiv changes to the codegen module, which pass mtex layers to the uniform system even when not needed. Since Sergey is demoing OpenSubdiv in a few days, I'll go with the easy fix for now which is just checking for NULL pointer, but this needs to be fixed properly at some point.
2015-08-07 16:16:23 +02:00
if (location == -1 || value == NULL)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
return;
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_ASSERT_NO_GL_ERRORS("Pre Uniform Vector");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
if (length == 1) glUniform1fvARB(location, arraysize, value);
else if (length == 2) glUniform2fvARB(location, arraysize, value);
else if (length == 3) glUniform3fvARB(location, arraysize, value);
else if (length == 4) glUniform4fvARB(location, arraysize, value);
else if (length == 9) glUniformMatrix3fvARB(location, arraysize, 0, value);
else if (length == 16) glUniformMatrix4fvARB(location, arraysize, 0, value);
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_ASSERT_NO_GL_ERRORS("Post Uniform Vector");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
void GPU_shader_uniform_vector_int(GPUShader *UNUSED(shader), int location, int length, int arraysize, const int *value)
{
if (location == -1)
return;
GPU_ASSERT_NO_GL_ERRORS("Pre Uniform Vector");
if (length == 1) glUniform1ivARB(location, arraysize, value);
else if (length == 2) glUniform2ivARB(location, arraysize, value);
else if (length == 3) glUniform3ivARB(location, arraysize, value);
else if (length == 4) glUniform4ivARB(location, arraysize, value);
GPU_ASSERT_NO_GL_ERRORS("Post Uniform Vector");
}
OpenGL: implemenation of fixed function lighting as per pixel GLSL shaders. The code is still unused, but the intention is to use this to solve the double sided lighting problem on NVidia, and to make the materials work on OpenGL ES 2.0 eventually. The code works and matches the fixed function lighting pretty much exactly, but still needs optimizations. The actual integration in object draw will be committed later when more fixing & testing, there's lots of different combinations and unclear OpenGL state here.
2013-02-26 00:49:42 +00:00
void GPU_shader_uniform_int(GPUShader *UNUSED(shader), int location, int value)
{
if (location == -1)
return;
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_CHECK_ERRORS_AROUND(glUniform1iARB(location, value));
OpenGL: implemenation of fixed function lighting as per pixel GLSL shaders. The code is still unused, but the intention is to use this to solve the double sided lighting problem on NVidia, and to make the materials work on OpenGL ES 2.0 eventually. The code works and matches the fixed function lighting pretty much exactly, but still needs optimizations. The actual integration in object draw will be committed later when more fixing & testing, there's lots of different combinations and unclear OpenGL state here.
2013-02-26 00:49:42 +00:00
}
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
void GPU_shader_geometry_stage_primitive_io(GPUShader *shader, int input, int output, int number)
{
glProgramParameteriEXT(shader->object, GL_GEOMETRY_INPUT_TYPE_EXT, input);
glProgramParameteriEXT(shader->object, GL_GEOMETRY_OUTPUT_TYPE_EXT, output);
glProgramParameteriEXT(shader->object, GL_GEOMETRY_VERTICES_OUT_EXT, number);
}
remove/tag unused args for view*.c, gpu*.c & image*.c
2010-10-15 12:29:02 +00:00
void GPU_shader_uniform_texture(GPUShader *UNUSED(shader), int location, GPUTexture *tex)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
GLenum arbnumber;
if (tex->number >= GG.maxtextures) {
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
fprintf(stderr, "Not enough texture slots.");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
return;
}
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (tex->number == -1)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
return;
style cleanup: follow style guide for formatting of if/for/while loops, and else if's
2012-03-24 06:38:07 +00:00
if (location == -1)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
return;
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_ASSERT_NO_GL_ERRORS("Pre Uniform Texture");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
arbnumber = (GLenum)((GLuint)GL_TEXTURE0_ARB + tex->number);
if (tex->number != 0) glActiveTextureARB(arbnumber);
Fix T40498 invalid textures flickering. Issue here is most likely sampler uniforms and textures not being updated properly when zero binding is created. Solution for now is to allow zero binding but when this happens use sexy pink invalid texture instead :p.
2014-06-10 01:36:53 +03:00
if (tex->bindcode != 0)
glBindTexture(tex->target, tex->bindcode);
else
GPU_invalid_tex_bind(tex->target);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
glUniform1iARB(location, tex->number);
glEnable(tex->target);
if (tex->number != 0) glActiveTextureARB(GL_TEXTURE0_ARB);
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_ASSERT_NO_GL_ERRORS("Post Uniform Texture");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
minor cleanup for string use - no functional changes - use more logical names for strings, noticed too many strings called `str` when reviewing name patch. - pass __func__ macro to uiBeginBlock(), quite a few names were wrong (copy/paste error).
2012-01-11 09:33:44 +00:00
int GPU_shader_get_attribute(GPUShader *shader, const char *name)
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
{
int index;
Debug GPU functionality from soc-viewport_fx by Jason Wilkins patch number D706 with changes: - WITH_GPU_DEBUG just creates a debug context (and enables the debug messaging system functions) but leaves the checks we had intact. Old patch added the debug functionality only if we had the flag on to save some performance. Rationale here is that we might not want to recompile blender just to get the extra information, and having users start blender with a -d flag to get the extra information is also useful for bug reports. Those checks already existed and most expensive ones are hidden behind a debug mode check so performance should not be that bad. - Did some cleanup of existing functionality: When things go wrong blender side, just print the error, don't check for GL errors first. - Did not port changes needed for GLES to regular glew.h - Got rid of duplicate or very similar new functionality. Generally, code is more moving things around/cleanup and should work exactly as before apart from the debug context, so it's safe to add even now. It also provides a nice substitute function for glu error descriptions
2015-02-16 21:19:12 +01:00
GPU_CHECK_ERRORS_AROUND(index = glGetAttribLocationARB(shader->object, name));
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
return index;
}
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
GPUShader *GPU_shader_get_builtin_shader(GPUBuiltinShader shader)
{
GPUShader *retval = NULL;
style cleanup: whitespace, braces
2012-05-01 20:08:23 +00:00
switch (shader) {
case GPU_SHADER_VSM_STORE:
if (!GG.shaders.vsm_store)
Fix high quality depth of field on the Mac. Quite a few things wrong here: * Mac did not support EXT_draw_instanced, only ARB_draw_instanced * Draw instanced did not work unless data came from vertex buffer, which is second time we see weird things with vertex arrays in mac * There were a few stupid mistakes by me as well, such as binding to uniform locations for the wrong shaders (it's a wonder it ever worked :p)
2015-03-20 15:26:13 +01:00
GG.shaders.vsm_store = GPU_shader_create(datatoc_gpu_shader_vsm_store_vert_glsl, datatoc_gpu_shader_vsm_store_frag_glsl, NULL, NULL, NULL, 0, 0, 0);
style cleanup: whitespace, braces
2012-05-01 20:08:23 +00:00
retval = GG.shaders.vsm_store;
break;
case GPU_SHADER_SEP_GAUSSIAN_BLUR:
if (!GG.shaders.sep_gaussian_blur)
Fix high quality depth of field on the Mac. Quite a few things wrong here: * Mac did not support EXT_draw_instanced, only ARB_draw_instanced * Draw instanced did not work unless data came from vertex buffer, which is second time we see weird things with vertex arrays in mac * There were a few stupid mistakes by me as well, such as binding to uniform locations for the wrong shaders (it's a wonder it ever worked :p)
2015-03-20 15:26:13 +01:00
GG.shaders.sep_gaussian_blur = GPU_shader_create(datatoc_gpu_shader_sep_gaussian_blur_vert_glsl, datatoc_gpu_shader_sep_gaussian_blur_frag_glsl, NULL, NULL, NULL, 0, 0, 0);
style cleanup: whitespace, braces
2012-05-01 20:08:23 +00:00
retval = GG.shaders.sep_gaussian_blur;
break;
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
}
if (retval == NULL)
Cleanup: fix mismatch in printf formating (int/unsigned int). Noisy and annoying with new gcc5...
2015-06-19 12:29:06 +02:00
printf("Unable to create a GPUShader for builtin shader: %u\n", shader);
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
return retval;
}
Fix T31546 fragment program gets created every frame That was really crappy indeed. Now we have a separate API for low level OpenGL programs, plus a nice interface for GPU, also removes some GL calls from main code as a plus :) The source for the programs is also moved to nice external .glsl files (not sure which extension convention GPU assemply uses)
2015-04-09 20:20:33 +02:00
GPUProgram *GPU_shader_get_builtin_program(GPUBuiltinProgram program)
{
GPUProgram *retval = NULL;
switch (program) {
case GPU_PROGRAM_SMOKE:
if (!GG.shaders.smoke)
GG.shaders.smoke = GPU_program_shader_create(GPU_PROGRAM_TYPE_FRAGMENT, datatoc_gpu_program_smoke_frag_glsl);
retval = GG.shaders.smoke;
break;
case GPU_PROGRAM_SMOKE_COLORED:
if (!GG.shaders.smoke_colored)
GG.shaders.smoke_colored = GPU_program_shader_create(GPU_PROGRAM_TYPE_FRAGMENT, datatoc_gpu_program_smoke_color_frag_glsl);
retval = GG.shaders.smoke_colored;
break;
}
if (retval == NULL)
Cleanup: fix mismatch in printf formating (int/unsigned int). Noisy and annoying with new gcc5...
2015-06-19 12:29:06 +02:00
printf("Unable to create a GPUProgram for builtin program: %u\n", program);
Fix T31546 fragment program gets created every frame That was really crappy indeed. Now we have a separate API for low level OpenGL programs, plus a nice interface for GPU, also removes some GL calls from main code as a plus :) The source for the programs is also moved to nice external .glsl files (not sure which extension convention GPU assemply uses)
2015-04-09 20:20:33 +02:00
return retval;
}
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
#define MAX_DEFINES 100
GPUShader *GPU_shader_get_builtin_fx_shader(int effects, bool persp)
{
int offset;
char defines[MAX_DEFINES] = "";
/* avoid shaders out of range */
if (effects >= MAX_FX_SHADERS)
return NULL;
offset = 2 * effects;
if (persp) {
offset += 1;
strcat(defines, "#define PERSP_MATRIX\n");
}
if (!GG.shaders.fx_shaders[offset]) {
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
GPUShader *shader;
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
switch (effects) {
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
case GPU_SHADER_FX_SSAO:
Fix high quality depth of field on the Mac. Quite a few things wrong here: * Mac did not support EXT_draw_instanced, only ARB_draw_instanced * Draw instanced did not work unless data came from vertex buffer, which is second time we see weird things with vertex arrays in mac * There were a few stupid mistakes by me as well, such as binding to uniform locations for the wrong shaders (it's a wonder it ever worked :p)
2015-03-20 15:26:13 +01:00
GG.shaders.fx_shaders[offset] = GPU_shader_create(datatoc_gpu_shader_fx_vert_glsl, datatoc_gpu_shader_fx_ssao_frag_glsl, NULL, datatoc_gpu_shader_fx_lib_glsl, defines, 0, 0, 0);
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
break;
case GPU_SHADER_FX_DEPTH_OF_FIELD_PASS_ONE:
strcat(defines, "#define FIRST_PASS\n");
Fix high quality depth of field on the Mac. Quite a few things wrong here: * Mac did not support EXT_draw_instanced, only ARB_draw_instanced * Draw instanced did not work unless data came from vertex buffer, which is second time we see weird things with vertex arrays in mac * There were a few stupid mistakes by me as well, such as binding to uniform locations for the wrong shaders (it's a wonder it ever worked :p)
2015-03-20 15:26:13 +01:00
GG.shaders.fx_shaders[offset] = GPU_shader_create(datatoc_gpu_shader_fx_dof_vert_glsl, datatoc_gpu_shader_fx_dof_frag_glsl, NULL, datatoc_gpu_shader_fx_lib_glsl, defines, 0, 0, 0);
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
break;
case GPU_SHADER_FX_DEPTH_OF_FIELD_PASS_TWO:
strcat(defines, "#define SECOND_PASS\n");
Fix high quality depth of field on the Mac. Quite a few things wrong here: * Mac did not support EXT_draw_instanced, only ARB_draw_instanced * Draw instanced did not work unless data came from vertex buffer, which is second time we see weird things with vertex arrays in mac * There were a few stupid mistakes by me as well, such as binding to uniform locations for the wrong shaders (it's a wonder it ever worked :p)
2015-03-20 15:26:13 +01:00
GG.shaders.fx_shaders[offset] = GPU_shader_create(datatoc_gpu_shader_fx_dof_vert_glsl, datatoc_gpu_shader_fx_dof_frag_glsl, NULL, datatoc_gpu_shader_fx_lib_glsl, defines, 0, 0, 0);
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
break;
case GPU_SHADER_FX_DEPTH_OF_FIELD_PASS_THREE:
strcat(defines, "#define THIRD_PASS\n");
Fix high quality depth of field on the Mac. Quite a few things wrong here: * Mac did not support EXT_draw_instanced, only ARB_draw_instanced * Draw instanced did not work unless data came from vertex buffer, which is second time we see weird things with vertex arrays in mac * There were a few stupid mistakes by me as well, such as binding to uniform locations for the wrong shaders (it's a wonder it ever worked :p)
2015-03-20 15:26:13 +01:00
GG.shaders.fx_shaders[offset] = GPU_shader_create(datatoc_gpu_shader_fx_dof_vert_glsl, datatoc_gpu_shader_fx_dof_frag_glsl, NULL, datatoc_gpu_shader_fx_lib_glsl, defines, 0, 0, 0);
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
break;
case GPU_SHADER_FX_DEPTH_OF_FIELD_PASS_FOUR:
strcat(defines, "#define FOURTH_PASS\n");
Fix high quality depth of field on the Mac. Quite a few things wrong here: * Mac did not support EXT_draw_instanced, only ARB_draw_instanced * Draw instanced did not work unless data came from vertex buffer, which is second time we see weird things with vertex arrays in mac * There were a few stupid mistakes by me as well, such as binding to uniform locations for the wrong shaders (it's a wonder it ever worked :p)
2015-03-20 15:26:13 +01:00
GG.shaders.fx_shaders[offset] = GPU_shader_create(datatoc_gpu_shader_fx_dof_vert_glsl, datatoc_gpu_shader_fx_dof_frag_glsl, NULL, datatoc_gpu_shader_fx_lib_glsl, defines, 0, 0, 0);
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
break;
case GPU_SHADER_FX_DEPTH_OF_FIELD_PASS_FIVE:
strcat(defines, "#define FIFTH_PASS\n");
Fix high quality depth of field on the Mac. Quite a few things wrong here: * Mac did not support EXT_draw_instanced, only ARB_draw_instanced * Draw instanced did not work unless data came from vertex buffer, which is second time we see weird things with vertex arrays in mac * There were a few stupid mistakes by me as well, such as binding to uniform locations for the wrong shaders (it's a wonder it ever worked :p)
2015-03-20 15:26:13 +01:00
GG.shaders.fx_shaders[offset] = GPU_shader_create(datatoc_gpu_shader_fx_dof_vert_glsl, datatoc_gpu_shader_fx_dof_frag_glsl, NULL, datatoc_gpu_shader_fx_lib_glsl, defines, 0, 0, 0);
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
break;
case GPU_SHADER_FX_DEPTH_OF_FIELD_HQ_PASS_ONE:
strcat(defines, "#define FIRST_PASS\n");
Fix high quality depth of field on the Mac. Quite a few things wrong here: * Mac did not support EXT_draw_instanced, only ARB_draw_instanced * Draw instanced did not work unless data came from vertex buffer, which is second time we see weird things with vertex arrays in mac * There were a few stupid mistakes by me as well, such as binding to uniform locations for the wrong shaders (it's a wonder it ever worked :p)
2015-03-20 15:26:13 +01:00
GG.shaders.fx_shaders[offset] = GPU_shader_create(datatoc_gpu_shader_fx_dof_hq_vert_glsl, datatoc_gpu_shader_fx_dof_hq_frag_glsl, NULL, datatoc_gpu_shader_fx_lib_glsl, defines, 0, 0, 0);
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
break;
case GPU_SHADER_FX_DEPTH_OF_FIELD_HQ_PASS_TWO:
strcat(defines, "#define SECOND_PASS\n");
Fix high quality depth of field on the Mac. Quite a few things wrong here: * Mac did not support EXT_draw_instanced, only ARB_draw_instanced * Draw instanced did not work unless data came from vertex buffer, which is second time we see weird things with vertex arrays in mac * There were a few stupid mistakes by me as well, such as binding to uniform locations for the wrong shaders (it's a wonder it ever worked :p)
2015-03-20 15:26:13 +01:00
shader = GPU_shader_create(datatoc_gpu_shader_fx_dof_hq_vert_glsl, datatoc_gpu_shader_fx_dof_hq_frag_glsl, datatoc_gpu_shader_fx_dof_hq_geo_glsl, datatoc_gpu_shader_fx_lib_glsl,
defines, GL_POINTS, GL_TRIANGLE_STRIP, 4);
GG.shaders.fx_shaders[offset] = shader;
Depth of field high quality: A new checkbox "High quality" is provided in camera settings to enable this. This creates a depth of field that is much closer to the rendered result and even supports aperture blades in the effect, but it's more expensive too. There are optimizations to do here since the technique is very fill rate heavy. People, be careful, this -can- lock up your screen if depth of field blurring is too extreme. Technical details: This uses geometry shaders + instancing and is an adaptation of techniques gathered from http://bartwronski.com/2014/04/07/bokeh-depth-of-field-going-insane- http://advances.realtimerendering.com/s2011/SousaSchulzKazyan%20- %20in%20Real-Time%20Rendering%20Course).ppt TODOs: * Support dithering to minimize banding. * Optimize fill rate in geometry shader.
2015-03-03 17:47:31 +01:00
break;
case GPU_SHADER_FX_DEPTH_OF_FIELD_HQ_PASS_THREE:
strcat(defines, "#define THIRD_PASS\n");
Fix high quality depth of field on the Mac. Quite a few things wrong here: * Mac did not support EXT_draw_instanced, only ARB_draw_instanced * Draw instanced did not work unless data came from vertex buffer, which is second time we see weird things with vertex arrays in mac * There were a few stupid mistakes by me as well, such as binding to uniform locations for the wrong shaders (it's a wonder it ever worked :p)
2015-03-20 15:26:13 +01:00
GG.shaders.fx_shaders[offset] = GPU_shader_create(datatoc_gpu_shader_fx_dof_hq_vert_glsl, datatoc_gpu_shader_fx_dof_hq_frag_glsl, NULL, datatoc_gpu_shader_fx_lib_glsl, defines, 0, 0, 0);
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
break;
Compositing works with X-Ray. Basically, before drawing X-Rays, we now bind a second depth buffer. After drawing XRays, we do an extra resolve pass where we overwrite the non-XRay depth buffer in pixels where the depth is not maximum (which means background pixel, since depth is cleared before drawing X-Ray objects). This ensures both scene and X-Rays keep their depth values and are ready for compositing. Well, the odd effect due to depth discontinuities can be expected, and X-Rays are a bit more expensive (extra buffer + resolve pass) but at least X-Rays won't invalidate depth values anymore. Whee!
2015-02-17 18:08:28 +01:00
case GPU_SHADER_FX_DEPTH_RESOLVE:
Fix high quality depth of field on the Mac. Quite a few things wrong here: * Mac did not support EXT_draw_instanced, only ARB_draw_instanced * Draw instanced did not work unless data came from vertex buffer, which is second time we see weird things with vertex arrays in mac * There were a few stupid mistakes by me as well, such as binding to uniform locations for the wrong shaders (it's a wonder it ever worked :p)
2015-03-20 15:26:13 +01:00
GG.shaders.fx_shaders[offset] = GPU_shader_create(datatoc_gpu_shader_fx_vert_glsl, datatoc_gpu_shader_fx_depth_resolve_glsl, NULL, NULL, defines, 0, 0, 0);
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
}
}
return GG.shaders.fx_shaders[offset];
}
style cleanup
2012-11-01 09:54:00 +00:00
void GPU_shader_free_builtin_shaders(void)
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
{
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
int i;
style cleanup: whitespace, braces
2012-05-01 20:08:23 +00:00
if (GG.shaders.vsm_store) {
Fix T31546 fragment program gets created every frame That was really crappy indeed. Now we have a separate API for low level OpenGL programs, plus a nice interface for GPU, also removes some GL calls from main code as a plus :) The source for the programs is also moved to nice external .glsl files (not sure which extension convention GPU assemply uses)
2015-04-09 20:20:33 +02:00
GPU_shader_free(GG.shaders.vsm_store);
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
GG.shaders.vsm_store = NULL;
}
style cleanup: whitespace, braces
2012-05-01 20:08:23 +00:00
if (GG.shaders.sep_gaussian_blur) {
Fix T31546 fragment program gets created every frame That was really crappy indeed. Now we have a separate API for low level OpenGL programs, plus a nice interface for GPU, also removes some GL calls from main code as a plus :) The source for the programs is also moved to nice external .glsl files (not sure which extension convention GPU assemply uses)
2015-04-09 20:20:33 +02:00
GPU_shader_free(GG.shaders.sep_gaussian_blur);
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
GG.shaders.sep_gaussian_blur = NULL;
}
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
Fix T31546 fragment program gets created every frame That was really crappy indeed. Now we have a separate API for low level OpenGL programs, plus a nice interface for GPU, also removes some GL calls from main code as a plus :) The source for the programs is also moved to nice external .glsl files (not sure which extension convention GPU assemply uses)
2015-04-09 20:20:33 +02:00
if (GG.shaders.smoke) {
GPU_program_free(GG.shaders.smoke);
GG.shaders.smoke = NULL;
}
if (GG.shaders.smoke_colored) {
GPU_program_free(GG.shaders.smoke_colored);
GG.shaders.smoke_colored = NULL;
}
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
for (i = 0; i < 2 * MAX_FX_SHADERS; i++) {
if (GG.shaders.fx_shaders[i]) {
Fix T31546 fragment program gets created every frame That was really crappy indeed. Now we have a separate API for low level OpenGL programs, plus a nice interface for GPU, also removes some GL calls from main code as a plus :) The source for the programs is also moved to nice external .glsl files (not sure which extension convention GPU assemply uses)
2015-04-09 20:20:33 +02:00
GPU_shader_free(GG.shaders.fx_shaders[i]);
Viewport compositing - first code This commit introduces a few ready made effects for the 3D viewport and OpenGL rendering. Included effects are Depth of Field, accessible from camera view and screen space ambient occlusion. Those effects can be turned on and tweaked from the shading panel in the 3D viewport. Off screen rendering will use the settings of the current camera. WIP documentation can be found here: http://wiki.blender.org/index.php/User:Psy-Fi/Framebuffer_Post-processing
2015-02-12 18:54:41 +01:00
GG.shaders.fx_shaders[i] = NULL;
}
}
Merging phase 1 of the BGE Harmony branch: * Shadow color now usable in the BGE * Simplified the shadow panel while "Blender Game" renderer is active * Added variance shadow maps for the BGE * Buffered shadows on sun lamps in the BGE (orthographic) * Light textures in the BGE
2012-05-01 02:50:17 +00:00
}
Add debug option --debug-gpumem to show GPU memory used in status bar. Only used in ATIs and NVIDIAs. Used extensions are: https://www.opengl.org/registry/specs/ATI/meminfo.txt http://developer.download.nvidia.com/opengl/specs/ If you read the documentation, the numbers are not supposed to be exact and also depend on the time when the call is made. The numbers can also change quite quickly. It's only meant to give a rough measure of what is going on.
2015-04-24 14:11:05 +02:00
bool GPU_mem_stats_supported(void)
{
return (GLEW_NVX_gpu_memory_info || (GLEW_ATI_meminfo)) && (G.debug & G_DEBUG_GPU_MEM);
}
void GPU_mem_stats_get(int *totalmem, int *freemem)
{
if (GLEW_NVX_gpu_memory_info) {
/* returned value in Kb */
glGetIntegerv(GL_GPU_MEMORY_INFO_TOTAL_AVAILABLE_MEMORY_NVX, totalmem);
glGetIntegerv(GL_GPU_MEMORY_INFO_CURRENT_AVAILABLE_VIDMEM_NVX, freemem);
}
else if (GLEW_ATI_meminfo) {
int stats[4];
glGetIntegerv(GL_TEXTURE_FREE_MEMORY_ATI, stats);
*freemem = stats[0];
*totalmem = 0;
}
else {
*totalmem = 0;
*freemem = 0;
}
}
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
#if 0 /* unused */
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
/* GPUPixelBuffer */
typedef struct GPUPixelBuffer {
GLuint bindcode[2];
GLuint current;
int datasize;
int numbuffers;
int halffloat;
} GPUPixelBuffer;
void GPU_pixelbuffer_free(GPUPixelBuffer *pb)
{
if (pb->bindcode[0])
glDeleteBuffersARB(pb->numbuffers, pb->bindcode);
MEM_freeN(pb);
}
GPUPixelBuffer *gpu_pixelbuffer_create(int x, int y, int halffloat, int numbuffers)
{
GPUPixelBuffer *pb;
if (!GLEW_ARB_multitexture || !GLEW_EXT_pixel_buffer_object)
return NULL;
pb = MEM_callocN(sizeof(GPUPixelBuffer), "GPUPBO");
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
pb->datasize = x * y * 4 * (halffloat ? 16 : 8);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
pb->numbuffers = numbuffers;
pb->halffloat = halffloat;
correct indentation and some whitespace edits (no functional changes)
2011-11-11 13:09:14 +00:00
glGenBuffersARB(pb->numbuffers, pb->bindcode);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
if (!pb->bindcode[0]) {
fprintf(stderr, "GPUPixelBuffer allocation failed\n");
GPU_pixelbuffer_free(pb);
return NULL;
}
return pb;
}
void GPU_pixelbuffer_texture(GPUTexture *tex, GPUPixelBuffer *pb)
{
void *pixels;
int i;
spaces -> tabs, (4 spaces == 1 tab, only for white space preceding text)
2010-03-22 09:30:00 +00:00
glBindTexture(GL_TEXTURE_RECTANGLE_EXT, tex->bindcode);
converted more mixed tab/space indentations to tabs. only whitespace changes.
2011-04-21 15:53:30 +00:00
for (i = 0; i < pb->numbuffers; i++) {
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
glBindBufferARB(GL_PIXEL_UNPACK_BUFFER_EXT, pb->bindcode[pb->current]);
glBufferDataARB(GL_PIXEL_UNPACK_BUFFER_EXT, pb->datasize, NULL,
converted more mixed tab/space indentations to tabs. only whitespace changes.
2011-04-21 15:53:30 +00:00
GL_STREAM_DRAW_ARB);
whitespace only, no functional change mixed tabs/spaces --> tabs.
2011-04-21 13:11:51 +00:00
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
pixels = glMapBufferARB(GL_PIXEL_UNPACK_BUFFER_EXT, GL_WRITE_ONLY);
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
# if 0
memcpy(pixels, _oImage.data(), pb->datasize);
# endif
whitespace only, no functional change mixed tabs/spaces --> tabs.
2011-04-21 13:11:51 +00:00
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
if (!glUnmapBufferARB(GL_PIXEL_UNPACK_BUFFER_EXT)) {
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
fprintf(stderr, "Could not unmap OpenGL PBO\n");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
break;
}
}
spaces -> tabs, (4 spaces == 1 tab, only for white space preceding text)
2010-03-22 09:30:00 +00:00
glBindTexture(GL_TEXTURE_RECTANGLE_EXT, 0);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
static int pixelbuffer_map_into_gpu(GLuint bindcode)
{
void *pixels;
spaces -> tabs, (4 spaces == 1 tab, only for white space preceding text)
2010-03-22 09:30:00 +00:00
glBindBufferARB(GL_PIXEL_UNPACK_BUFFER_EXT, bindcode);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
pixels = glMapBufferARB(GL_PIXEL_UNPACK_BUFFER_EXT, GL_WRITE_ONLY);
/* do stuff in pixels */
spaces -> tabs, (4 spaces == 1 tab, only for white space preceding text)
2010-03-22 09:30:00 +00:00
if (!glUnmapBufferARB(GL_PIXEL_UNPACK_BUFFER_EXT)) {
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
fprintf(stderr, "Could not unmap OpenGL PBO\n");
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
return 0;
spaces -> tabs, (4 spaces == 1 tab, only for white space preceding text)
2010-03-22 09:30:00 +00:00
}
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
return 1;
}
static void pixelbuffer_copy_to_texture(GPUTexture *tex, GPUPixelBuffer *pb, GLuint bindcode)
{
GLenum type = (pb->halffloat)? GL_HALF_FLOAT_NV: GL_UNSIGNED_BYTE;
spaces -> tabs, (4 spaces == 1 tab, only for white space preceding text)
2010-03-22 09:30:00 +00:00
glBindTexture(GL_TEXTURE_RECTANGLE_EXT, tex->bindcode);
glBindBufferARB(GL_PIXEL_UNPACK_BUFFER_EXT, bindcode);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
glTexSubImage2D(GL_TEXTURE_RECTANGLE_EXT, 0, 0, 0, tex->w, tex->h, GL_RGBA, type, NULL);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
glBindBufferARB(GL_PIXEL_UNPACK_BUFFER_EXT, 0);
spaces -> tabs, (4 spaces == 1 tab, only for white space preceding text)
2010-03-22 09:30:00 +00:00
glBindTexture(GL_TEXTURE_RECTANGLE_EXT, 0);
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
void GPU_pixelbuffer_async_to_gpu(GPUTexture *tex, GPUPixelBuffer *pb)
{
int newbuffer;
if (pb->numbuffers == 1) {
pixelbuffer_copy_to_texture(tex, pb, pb->bindcode[0]);
pixelbuffer_map_into_gpu(pb->bindcode[0]);
}
else {
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
pb->current = (pb->current + 1) % pb->numbuffers;
newbuffer = (pb->current + 1) % pb->numbuffers;
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
pixelbuffer_map_into_gpu(pb->bindcode[newbuffer]);
pixelbuffer_copy_to_texture(tex, pb, pb->bindcode[pb->current]);
spaces -> tabs, (4 spaces == 1 tab, only for white space preceding text)
2010-03-22 09:30:00 +00:00
}
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
}
cleanup: use spaces for alignment while studying GPU lib
2015-03-23 15:29:42 -04:00
#endif /* unused */
Merge of first part of changes from the apricot branch, especially the features that are needed to run the game. Compile tested with scons, make, but not cmake, that seems to have an issue not related to these changes. The changes include: * GLSL support in the viewport and game engine, enable in the game menu in textured draw mode. * Synced and merged part of the duplicated blender and gameengine/ gameplayer drawing code. * Further refactoring of game engine drawing code, especially mesh storage changed a lot. * Optimizations in game engine armatures to avoid recomputations. * A python function to get the framerate estimate in game. * An option take object color into account in materials. * An option to restrict shadow casters to a lamp's layers. * Increase from 10 to 18 texture slots for materials, lamps, word. An extra texture slot shows up once the last slot is used. * Memory limit for undo, not enabled by default yet because it needs the .B.blend to be changed. * Multiple undo for image painting. * An offset for dupligroups, so not all objects in a group have to be at the origin.
2008-09-04 20:51:28 +00:00
Copy Permalink