Originally we had 2DFilters (m_filtermanager) stored in RenderTools. That way filters were stored globally and were being called once per each scene. This was producing two big problems: (1) performance and (2) flexibility of use.
(1) Performance - To run the filters 2X == 2X slower
(2) flexibility of use - Very often we want the filter in the scene but not in the UI for example.
For those reasons I believe that 2DFilters with multiple scenes was very useless or unpredictable. I hope they work fine now.
To make it work as before (2.4) you can simply recreate the 2dfilter actuators across the scenes.
* * * * *
Imagine that we have:
(a) Main Scene
(b) Overlay Scene
in Main Scene the Z Buffer and RGB will be from the main scene.
in Overlay Scene the Z Buffer will be from the Overlay Scene and the RBG buffer is from both [(a + 2D Filter) + b].
So in pseudo code if we have a,b,c,d,e scenes we have: (2DFilterE(2DFilterD(2DFilterC(2DFilterB(2DFilterA(a) + b) + c) + d) + e)
The problem was: the Blender default camera has DOF distance as 0.0. Since we are using this as Focal Length for the stereo calculation we had terrible stereo by default.
Fix: whenever DOF == 0.0 we use focal length as eye separation * 30.0 (known to be a reasonable value)
Now the default eye separation value is 0.10 (reasonable for games with 1 meter == 1 B.U.
The focallength used is the camera focal length (DOF settings). It allow you to even use different focal lengths for different scenes (good for UI)
In order to change it you can change the camera focal length or use Rasterizer.setFocalLength.
If you use the Rasterizer method it will use this value for all the cameras.
ToDo:
- Blenderplayer settings
- Update wiki documentation (any volunteer)?
* Note to stereo fans:
I don't have a real stereo environment to test it (other than cheap cyan-red glasses). If you can give it a try in a more robust system and report bugs or problems with BGE current system please let me know. I would be glad to help to make it work 100% by the time Blender 2.5 is out.
For the record, BGE is using the method known as 'parallel axis asymmetric frustum perspective projection'. This method is well documented here:
http://local.wasp.uwa.edu.au/~pbourke/miscellaneous/stereographics/stereorender/
Even a static mesh can be used as replacement: the mesh
will be instantiated with the soft body settings of the
object. The position and orientation of the soft body
is preserved after the replacement.
Known limitation: the velocity of the soft body is reset
aftet the replacement. This is because soft body don't
have a well defined velocity.
GLEW update to version 1.5.1 [11-03-08]
this opens room for Geometry Shader support.
* - Brecht, Campbell told me you did some local changes in order to make it right in Linux. I get to you in order to know what those changes are (or feel free to commit them directly)
* bring back 'player' libtype, after investigation with ideasman.
scons/mingw works nicely, for some reason msvc fails to link still, will look further into it.
* further cleaning of 'player' stuff. Now only 3 libs are remaining, of which ideally the stubs lib will be fixed at some point, fading away into the dark history of not-so-nice code. The current blenderplayer part is still a little bit hackish, I'll see if I can find a better alternative, for now it works good enough.
* first working changes to get blenderplayer linking
* blenderplayer/ moved into source/ (CMakeLists.txt changed for that too)
* added externs for bprogname to gp_ghost, so that it links properly
* Add BGE_CXXFLAGS so we can get rid of hard-coded BGE compiler settings. This was only done for windows, but now linuxers and osxers should be able to set BGE-specific optimisation too. See the windows default configs for example.
Surprising this wasn't noticed before. Any mix of quads/tris caused the face verts of either quads/tries (whichever comes last).
Tested by exporting the KX_MeshProxy and re-importing as an OBJ.
This fix assumes there are only 2 m_darray's per face array which is currently true, but wont be if edge support is added back.
Add support back for reinstancePhysics mesh, a frequently requested feature in the BGE forums.
from what I can tell Sumo supported this but bullet never did.
Currently only accessible via python at the moment.
- rigid body, dynamic, static types work.
- instanced physics meshes are modified too.
- compound shapes are not supported.
Physics mesh can be re-instanced from...
* shape keys & armature deformations
* subsurf (any other modifiers too)
* RAS_TexVert's (can be modified from python)
Moved the reinstancePhysicsMesh functions from RAS_MeshObject into KX_GameObject since the physics data is stored here.
video and blend file demo.
http://www.graphicall.org/ftp/ideasman42/reinstance.ogvhttp://www.graphicall.org/ftp/ideasman42/reinstance_demo.blend
Logic Panel:
- world settings (moved from world)
... that includes physic engine selection + gravity
- game player (from gamesettings, it wasn't wrapped)
- stereo/dome (from gamesettings, it wasn't wrapped)
... separated stereom into stereoflag and stereomode
- properties
... (didn't touch it)
Buttons Game Panel:
(wip panel)
- Physics (moved from Logic Panel)
... it will be a datablock in the future (right Campbell ?)
- Material Physics (not currently implemented)
... a datablock link to the materials of an object + the dynamic physic variables
* NOTE:
in readfile.c::do_version I couldn't do if(scene->world). There is something wrong with scenes with an unlinked world. So so far we are ignoring the old values....
* ensure all SConscripts are ready for win64-vc (where necessary).
* ensure we have proper _DEBUG flag for Python when we're doing a debug build.
* some cleaning up of linking etc.
* ensure /EHsc is there for game engine modules.
* Missing changes to release/windows/installer
* Sequencer fixes in source/blender/src/seqaudio.c dont apply to 2.5
* brechts fix for #18855 r20763 wasnt merged, does this apply to 2.5?
svn merge https://svn.blender.org/svnroot/bf-blender/trunk/blender -r19820:HEAD
Notes:
* Game and sequencer RNA, and sequencer header are now out of date
a bit after changes in trunk.
* I didn't know how to port these bugfixes, most likely they are
not needed anymore.
* Fix "duplicate strip" always increase the user count for ipo.
* IPO pinning on sequencer strips was lost during Undo.
SCA_RandomActuator: The random generator was shared between replicas and not deleted. Added ref counting between replicas to allow deletion at the end.
KX_Camera: The scenegraph node was not deleted for temporary cameras (ImageMirror and shadow), causing 500 bytes leak per frame and per shadow light.
KX_GameActuator: Global dictionary buffer was not deleted after saving.
KX_MotionState: The motion state for compound child was not deleted
KX_ReplaceMeshActuator: The mesh was unnecessarily converted for each actuator and not deleted, causing large memleak.
After these fix, YoFrankie runs without memleak.
This commit completes the support for modifiers in the BGE.
- The physic shape is generated according to the derived mesh.
This is true for all types of shapes and all types of
objects except soft body.
- Optimization for static derived mesh (mesh with modifiers
but no armature and no shape keys). Replicas will share
the derived mesh and the display list: less memory and
faster rendering. With this optimization, the static
derived mesh will render as fast as if the modifiers were
applied.
Known Limits:
- Sharing of mesh and display list is only possible between
in-game replicas or dupligroup. If you want to instantiate
multiple objects with modifiers, use dupligroup to ensure
best memory and GPU utilization.
- rayCast() will interact with the derived mesh as follow:
Hit position and hit normal are the real values according
to the derived mesh but the KX_PolyProxy object refers to
the original mesh. You should use it only to retrieve the
material.
- Dynamic derived mesh have very poor performance:
They use direct openGL calls for rendering (no support
for display list and vertex array) and they dont't share
the derived mesh memory. Always apply modifiers on dynamic
mesh for best performance.
- Time dependent modifiers are not supported.
- Modifiers are not supported for Bullet soft body.
This commit extends the technique of dynamic linked list to the logic
system to eliminate as much as possible temporaries, map lookup or
full scan. The logic engine is now free of memory allocation, which is
an important stability factor.
The overhead of the logic system is reduced by a factor between 3 and 6
depending on the logic setup. This is the speed-up you can expect on
a logic setup using simple bricks. Heavy bricks like python controllers
and ray sensors will still take about the same time to execute so the
speed up will be less important.
The core of the logic engine has been much reworked but the functionality
is still the same except for one thing: the priority system on the
execution of controllers. The exact same remark applies to actuators but
I'll explain for controllers only:
Previously, it was possible, with the "executePriority" attribute to set
a controller to run before any other controllers in the game. Other than
that, the sequential execution of controllers, as defined in Blender was
guaranteed by default.
With the new system, the sequential execution of controllers is still
guaranteed but only within the controllers of one object. the user can
no longer set a controller to run before any other controllers in the
game. The "executePriority" attribute controls the execution of controllers
within one object. The priority is a small number starting from 0 for the
first controller and incrementing for each controller.
If this missing feature is a must, a special method can be implemented
to set a controller to run before all other controllers.
Other improvements:
- Systematic use of reference in parameter passing to avoid unnecessary data copy
- Use pre increment in iterator instead of post increment to avoid temporary allocation
- Use const char* instead of STR_String whenever possible to avoid temporary allocation
- Fix reference counting bugs (memory leak)
- Fix a crash in certain cases of state switching and object deletion
- Minor speed up in property sensor
- Removal of objects during the game is a lot faster
This commit extend the technique of dynamic linked list to the mesh
slots so as to eliminate dumb scan or map lookup. It provides massive
performance improvement in the culling and in the rasterizer when
the majority of objects are static.
Other improvements:
- Compute the opengl matrix only for objects that are visible.
- Simplify hash function for GEN_HasedPtr
- Scan light list instead of general object list to render shadows
- Remove redundant opengl calls to set specularity, shinyness and diffuse
between each mesh slots.
- Cache GPU material to avoid frequent call to GPU_material_from_blender
- Only set once the fixed elements of mesh slot
- Use more inline function
The following table shows the performance increase between 2.48, 1st round
and this round of improvement. The test was done with a scene containing
40000 objects, of which 1000 are in the view frustrum approximately. The
object are simple textured cube to make sure the GPU is not the bottleneck.
As some of the rasterizer processing time has moved under culling, I present
the sum of scenegraph(includes culling)+rasterizer time
Scenegraph+rasterizer(ms) 2.48 1st round 3rd round
All objects static, 323.0 86.0 7.2
all visible, 1000 in
the view frustrum
All objects static, 219.0 49.7 N/A(*)
all invisible.
All objects moving, 323.0 105.6 34.7
all visible, 1000 in
the view frustrum
Scene destruction 40min 40min 4s
(*) : this time is not representative because the frame rate was at 60fps.
In that case, the GPU holds down the GE by frame sync. By design, the
overhead of the rasterizer is 0 when the the objects are invisible.
This table shows a global speed up between 9x and 45x compared to 2.48a
for scenegraph, culling and rasterizer overhead. The speed up goes much
higher when objects are invisible.
An additional 2-4x speed up is possible in the scenegraph by upgrading
the Moto library to use Eigen2 BLAS library instead of C++ classes but
the scenegraph is already so fast that it is not a priority right now.
Next speed up in logic: many things to do there...
