Really bad issue which meant code could fetch an image buffer from the
stored cache and modify it. Generally sequence image buffers could come
from the cache and should not be modified directly. Easily solved by
scaling a copy of the original.
Couple of issues:
- Fist/last frame calculation was wrong
- Keyframe selection might silently fail leading to unpredictable math
errors all over the place. Now if keyframe selection fails solver wouldn't
run.
Forbid OSL from polluting current conext with obscure stuff from
windows.h, it's not useful and unhealthy anyway.
Maybe we sohuld also forbid using abbreviated Glog constants as
well tho.
Buffers should actually be cleared before running operations on them,
but this doesn't work for some reason.
Note also that the sunbeams node can show some creases and hard aliasing
when the source point is close to a bright area with strong gradient.
To fix this a better filtering algorithm, dithering or ray sampling
would need to be implemented. In the meantime simply blurring the
sunbeams result a bit should help (or simply avoid putting the source on
a bright spot).
Default configuration already includes libmv and glog, so all
the dependencies are met. And logging is nice to have anyway.
This wouldn't give any slowdown, because all the logging is
silenced unless '--debug-cycles', plus no logging is done
from the speed critical code.
Since the aligned allocation of shader closures in OSL memory pool
this workaround is no longer needed.
Also put a comment which describes the desired layout of the structure
so array of shader closures is all nicely aligned.
This solves bugs like T42210 which are caused by compiler being
smart and using some SSE instructions to operate with closure
classes, which was failing because those classes are not allocated
by the regular allocator but allocated in memory pool in OSL.
With newer versions of OSL it is now possible to force closure
classes being aligned to a given boundary and this commit uses
this new functionality.
Unfortunately, it means we're no longer compatible with older
versions of OSL, only latest git version from upstream and our
branch at github are supported:
https://github.com/Nazg-Gul/OpenShadingLanguage/tree/blender-fixes
For OSX and Windows it's not an issue because libraries are
already updated there, Linux users would need to run install_deps
script.
Fix for T41294.
Rigid body constraints are not deleted, if the corresponding game objects are deleted.
Reviewers: moguri
Differential Revision: https://developer.blender.org/D701
after the completion of the action in "Flipper", layer is removed and the actuator mistakenly receive zero when trying to get the current frame
Patch Author: avrprj
Reviewers: moguri
Projects: #game_logic
Differential Revision: https://developer.blender.org/D906
This ensures proper values of currently running tasks in the pool
(previously difference between mutex locks when acquiring new job
and releasing it might in theory give wrong values).
OpenGL in background mode is not used, so we can skip mutex lock and filling in
the list which later is never used.
This gives unmeasurable speedup by skipping mutex lock, plus solves memory leak
in the background mode.
Thanks to an anonymous tip (or shall we say, a tip from "Anonymous" - thank you
whoever you are :) it is now possible to render out Grease Pencil shots from
the viewport with correct colours again! This has been broken for a few releases
now, so it's great that this works again now, completing the last part of the
pipeline again.
This helps to reduce jaggies from thin lines, while also resulting in nicer
lines elsewhere.
I'm not sure if it's just me, but it seems to render slightly differently to
before for 3d strokes too (i.e. they seem a bit softer). Hopefully the
difference isn't big enough to affect/degrade the art style of any projects.
DNA type.
The update function is for the Paint struct, containing a Brush pointer
property, not Brush itself. Probably went fine so far because was only
used for a notifier pointer, but still ...
Inspired by the previous commit, I've finally found a way to fix a long standing
limitation/bug which meant that Grease Pencil strokes in the Image Editor could
not be drawn using the fancy stroke tesselation code, and were instead done using
the plain old OpenGL strokes instead.
This needed a correction factor (currently hardcoded to be 1 / 1000)
as it seems that the image editor uses 1 unit as its MAXIMUM dimension
whereas everything else uses 1 unit = 1 pixel.
Collada does not support Bone length. It has only Joints.
This patch tries to improve the necessary "guesswork" for
importing Rigs.
Differential Revision: https://developer.blender.org/D920
* Same issue as T42760 was also reproducible in some cases in 3DView.
* You got an error message about missing RNA prop in some drop-called operators.
* You could not drop a movie file in nodes, (some cases of) 3DView, nor ImageEditor.
For now, let's move the GPencil pies to avoid conflicts with dyntopo and other things
* DQ = Main pie (previously Ctrl-D; DD was too unpredictable)
* DW = Settings pie
I'd kept the code around in the codebase until after the merge back to master
to avoid having too many conflicts if things changed there (or in case we
needed to roll back). Now, it's safe to jettison this!
This merge-commit brings in a number of new features and workflow/UI improvements for
working with Grease Pencil. While these were originally targetted at improving
the workflow for creating 3D storyboards in Blender using the Grease Pencil,
many of these changes should also prove useful in other workflows too.
The main highlights here are:
1) It is now possible to edit Grease Pencil strokes
- Use D Tab, or toggle the "Enable Editing" toggles in the Toolbar/Properties regions
to enter "Stroke Edit Mode". In this mode, many common editing tools will
operate on Grease Pencil stroke points instead.
- Tools implemented include Select, Select All/Border/Circle/Linked/More/Less,
Grab, Rotate, Scale, Bend, Shear, To Sphere, Mirror, Duplicate, Delete.
- Proportional Editing works when using the transform tools
2) Grease Pencil stroke settings can now be animated
NOTE: Currently drivers don't work, but if time allows, this may still be
added before the release.
3) Strokes can be drawn with "filled" interiors, using a separate set of
colour/opacity settings to the ones used for the lines themselves.
This makes use of OpenGL filled polys, which has the limitation of only
being able to fill convex shapes. Some artifacts may be visible on concave
shapes (e.g. pacman's mouth will be overdrawn)
4) "Volumetric Strokes" - An alternative drawing technique for stroke drawing
has been added which draws strokes as a series of screen-aligned discs.
While this was originally a partial experimental technique at getting better
quality 3D lines, the effects possible using this technique were interesting
enough to warrant making this a dedicated feature. Best results when partial
opacity and large stroke widths are used.
5) Improved Onion Skinning Support
- Different colours can be selected for the before/after ghosts. To do so,
enable the "colour wheel" toggle beside the Onion Skinning toggle, and set
the colours accordingly.
- Different numbers of ghosts can be shown before/after the current frame
6) Grease Pencil datablocks are now attached to the scene by default instead of
the active object.
- For a long time, the object-attachment has proved to be quite problematic
for users to keep track of. Now that this is done at scene level, it is
easier for most users to use.
- An exception for old files (and for any addons which may benefit from object
attachment instead), is that if the active object has a Grease Pencil datablock,
that will be used instead.
- It is not currently possible to choose object-attachment from the UI, but
it is simple to do this from the console instead, by doing:
context.active_object.grease_pencil = bpy.data.grease_pencil["blah"]
7) Various UI Cleanups
- The layers UI has been cleaned up to use a list instead of the nested-panels
design. Apart from saving space, this is also much nicer to look at now.
- The UI code is now all defined in Python. To support this, it has been necessary
to add some new context properties to make it easier to access these settings.
e.g. "gpencil_data" for the datablock
"active_gpencil_layer" and "active_gpencil_frame" for active data,
"editable_gpencil_strokes" for the strokes that can be edited
- The "stroke placement/alignment" settings (previously "Drawing Settings" at the
bottom of the Grease Pencil panel in the Properties Region) is now located in
the toolbar. These were more toolsettings than properties for how GPencil got drawn.
- "Use Sketching Sessions" has been renamed "Continuous Drawing", as per a
suggestion for an earlier discussion on developer.blender.org
- By default, the painting operator will wait for a mouse button to be pressed
before it starts creating the stroke. This is to make it easier to include
this operator in various toolbars/menus/etc. To get it immediately starting
(as when you hold down DKEy to draw), set "wait_for_input" to False.
- GPencil Layers can be rearranged in the "Grease Pencil" mode of the Action Editor
- Toolbar panels have been added to all the other editors which support these.
8) Pie menus for quick-access to tools
A set of experimental pie menus has been included for quick access to many
tools and settings. It is not necessary to use these to get things done,
but they have been designed to help make certain common tasks easier.
- Ctrl-D = The main pie menu. Reveals tools in a context sensitive and
spatially stable manner.
- D Q = "Quick Settings" pie. This allows quick access to the active
layer's settings. Notably, colours, thickness, and turning
onion skinning on/off.
Basically this commit gets rid of most of the derived mesh immediate mode
drawing (cases such as subsurf excluded). Even when VBO is turned off
in user preferences, we still use vertex arrays, which are very similar to
VBOs but memory is client side. Vertex arrays are OpenGL 1.1 so compatibility
is not an issue here.
Reviewers: campbellbarton, sergey, jwilkins
Differential Revision: https://developer.blender.org/D919
Basically, our drawing code assumed we always use the edit mesh
materials, which can be different from the derived mesh
materials in modifiers doing overrides. We usually we want to use the
derived mesh when it is available instead.
There are two fixes here for both solid and textured mode. Unfortunately
the fixes do not help to make the display code less labyrinthian but I
expect this "should work" (tm and famous last words)
Solid mode fix is 95% from Bastien, thanks!
This is added in the spirit of the general cycles GLSL system
which is pretty much WIP still.
This will only work on cycles at the moment but generating for blender
internal is possible too of course though it will be done in a separate
commit.
This hasn't been tested with all and every node in cycles, but
environment and regular textures with texture coordinates work.
There is some difference between the way cycles treats some coordinates,
which is in world space and the way GLSL treats them, which is in view
space.
We might want to explore and improve this further in the future.
...also </drumroll>
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)
Fixes the problem that for big sequences too many file handles were open at the same time.
Changes the playback handles that the audio sequencing code manages to be closed and reopened when needed. The metric used is the current playback position in relation to the strip. If the strip is more than 10 seconds (configurable) away from the playback cursor, the handle is released and reopened when needed.
See D915.
This patch includes the work done in the terrible consequencer branch
that hasn't been merged to master minus a few controversial and WIP
stuff, like strip parenting, new sequence data structs and cuddly
widgets.
What is included:
* Strip extensions only when slipping. It can very easily be made an
option but with a few strips with overlapping durations it makes view
too crowded and difficult to make out.
* Threaded waveform loading + code that restores waveforms on undo (not
used though, since sound_load recreates everything. There's a patch for
review D876)
* Toggle to enable backdrop in the strip sequence editor
* Toggle to easily turn on/off waveform display
* Snapping during transform on sequence boundaries. Snapping to start or
end of selection depends on position of mouse when invoking the operator
* Snapping of timeline indicator in sequencer to strip boundaries. To
use just press and hold ctrl while dragging.
Reviewers: campbellbarton
Differential Revision: https://developer.blender.org/D904
The issue was noticed with gcc-4.7 (used by the release build environment)
which didn't generate optimal enough code for BVH references swap. Seems it
looked up for the assign operator for each of the reference structure members
even though nothing special was required for assignment.
Forcing compiler to use simple memory copy gives speedup of like 2-3 times.
The issue doesn't happen with OSX's clang and new gcc-4.9, but since we're
gonna to stick to gcc-4.7 for official releases for quite some time still it's
nice to have performance issues resolved for all the compilers.
Didn't put the code into #ifdef so if in the future some issues appears with
alignment or assignment which need to happen as an operator we notice this
earlier.
Previosuly, world was shown on the background if "Render Only" was used.
Now user should be able to set it independently. This is a prelude to
(drumroll)...
It may be a bit early to do such cleanups, there are more important things
to do first, but those things are extremely time consuming and I wanted to
get rid of them first ;)
* Alpha Property was removed (Fix T42690)
* Some tweaks to make the panel look better again.
* Use abreviated form "Multiple Importance" everywhere, for consistency.
gpencil_data_duplicate() was being used for gp drawing undo buffers, where using an
exact copy is exactly what we want/need. Instead, the code here now has an additional
arg for determining whether a direct copy is warranted or not.
Textures attached to particle systems are now get their animation data listed
under the particle systems they are attached to now. This is the most convenient
and direct way that these can get included
This way we can have scheduler capable of scheduling tasks on all the CPUs
but in the same time we can limit tasks like baking (in the future) to use
no more than given number of threads.
Basically, get the grid increments and reuse them when snapping. System
is slightly crappy here, we should calculate those factors only once,
but leaving as todo for later.
Basically, `angle_compat_rad()` was completely broken -
example of result it could produce:
| new angle | compat angle | result
| -0.000000 | 3.141593 | -> 3.141593
... Where 0.0 (or 2 * PI) would be expected!
In the process, I've removed the old operator (ANIM_OT_channels_visibility_set)
and folded that option in with the hide operator, to make this consistent
with how this is done in the 3D view and other parts of Blender.
Now the hotkeys here work in line with what's done for other parts of Blender
* H = Hide selected
* Shift-H = Hide unselected (i.e. old VKEY behaviour)
* Alt-H = Reveal all
Revised the tools for managing which FCurves are visible in the Graph Editor
curves area. Now, there are the following tools in place:
* V (channels region only) = Hide all curves except those in selected channels [OLD]
* H = Hide all selected curves [NEW]
* Shift-H = Show all previously hidden curves [NEW]
I've removed the old operator to toggle visibility status of selected curves,
as it doesn't seem that useful anymore.
Along with some minor cleanup and simplifications.
Reviewers: campbellbarton
Subscribers: sergey
Differential Revision: https://developer.blender.org/D903
* Alpha Property was removed (Fix T42690)
* Some tweaks to make the panel look better again.
* Use abreviated form "Multiple Importance" everywhere, for consistency.
gpencil_data_duplicate() was being used for gp drawing undo buffers, where using an
exact copy is exactly what we want/need. Instead, the code here now has an additional
arg for determining whether a direct copy is warranted or not.
Textures attached to particle systems are now get their animation data listed
under the particle systems they are attached to now. This is the most convenient
and direct way that these can get included
This way we can have scheduler capable of scheduling tasks on all the CPUs
but in the same time we can limit tasks like baking (in the future) to use
no more than given number of threads.
Basically, get the grid increments and reuse them when snapping. System
is slightly crappy here, we should calculate those factors only once,
but leaving as todo for later.
Basically, `angle_compat_rad()` was completely broken -
example of result it could produce:
| new angle | compat angle | result
| -0.000000 | 3.141593 | -> 3.141593
... Where 0.0 (or 2 * PI) would be expected!
In the process, I've removed the old operator (ANIM_OT_channels_visibility_set)
and folded that option in with the hide operator, to make this consistent
with how this is done in the 3D view and other parts of Blender.
Now the hotkeys here work in line with what's done for other parts of Blender
* H = Hide selected
* Shift-H = Hide unselected (i.e. old VKEY behaviour)
* Alt-H = Reveal all
Revised the tools for managing which FCurves are visible in the Graph Editor
curves area. Now, there are the following tools in place:
* V (channels region only) = Hide all curves except those in selected channels [OLD]
* H = Hide all selected curves [NEW]
* Shift-H = Show all previously hidden curves [NEW]
I've removed the old operator to toggle visibility status of selected curves,
as it doesn't seem that useful anymore.
Along with some minor cleanup and simplifications.
Reviewers: campbellbarton
Subscribers: sergey
Differential Revision: https://developer.blender.org/D903
So now cases when object has both hair motion blur and deformation motion blur
vector pass is all correct.
We could get rid of the flag in the future, still need to look deeper into all
the areas trying to find a more clear solution.
The issue was caused by the threading conflict between main thread which
might free environment map and render thread which might be using that
environment map.
Solved by stopping preview jobs when changing environment map.
Sure there are still ways to run into threading conflict, but would rather
solve this things case-by-case.
Issue was caused by mismatch in pre/post transform matrix spaces for mesh and
curve vectors. This happened because of current way how static transform apply
works: it only stores post/pre in the world space if there's triangle motion
exists. This lead to situation when there's no triangle motion happening but
was hair motion happening.
After long time of trying to solve it in a nice way, ended up solving it in
a bit slow way -- pre/post transform is still storing in the same spaces as
they used to be stored and just convert hair pre/post position to a world
space in the kernel.
This is because currently it's not so clear how to deal with cases when curve
and mesh motion needs different space of pre/post transform (which happens in
cases when only one of the motions exists).
Would think of some magic, and meanwhile artists could be happy with proper
render results.
* 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).
The issue was caused by render engine duplicating the curve object because
it then converts the object to mesh. The mesh deform duplication code didn't
duplicate binded data, so after duplication modifier is no longer applyable.
So now copyData of mesh deform would copy data needed for binding. This solves
reported render bug and also solves possible frustration when duplicating an
object with mesh deform in the viewport with Shift-D.
Checked other modifiers and laplacian deform already was copying binded data.
Didn't see other modifiers which might also need to copy extra data.
It may be a bit early to do such cleanups, there are more important things
to do first, but those things are extremely time consuming and I wanted to
get rid of them first ;)
Patch by Martin Vykoukal, thanks!
This patch adds ability to change brush parameters with keyboard, which
is missing functionality from 2.4x.
Original report: T28811
Reviewers: psy-fi
Differential Revision: https://developer.blender.org/D837
This is mainly to address old issue when one need to have SDL library installed
in order to use our official builds. Some hip distros already installs SDL,
but it's not quite the same across all the variety of the distros.
We also now switching to SDL-2.0, most of the distros have it in repositories
already, so it shouldn't be huge deal to install it if needed.
Reviewers: campbellbarton
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D878
Mainly:
* Use 'for' loops instead of 'while' ones (saves many lines and regroup most loop handling on one line).
* Use float[3] pointers where possible.
Based on investigation by sergey (Sergey Sharybin) and revzin (Grigory Revzin).
Based on patch D460 by revzin (Grigory Revzin).
Differential Revision: https://developer.blender.org/D460
Thing is, those functions always reallocate the whole keyblock's data mem,
while in some cases we already have right amount of elements, so we can just
copy over. Further more, `BKE_key_convert_from_offset`, despite its name,
was not making any check nor allocation on keyblock's data elements!
So split 'copy' operation itself in `BKE_key_update_from_...`,
where no mem checks/operations are performed (only an assert).
Only useful in sculpt mode currently, but will be used by fix for T35170 too.
Rather straightforward, allows for 'DATA' icons (like mat or tex 'previews')
to be used as icon for operator button or items of an enum.
Patch by Simon Lušenc, with minor cleanup by self.
This can be used to avoid redundant looping when we only want to know if a list is smaller then some size.
also remove paranoid NULL check in list counting.
For FCurves where all the keyframes use the "simple" interpolation types
(i.e. Constant, Linear, and Bezier), we now use the old FCurve drawing
code that was used prior to the Easing Equations changes. This should
be generally faster in general.
When the "High Quality Line Drawing" option (View menu) is disabled,
the sampling rate (i.e. the size of timesteps to use when sampling
the FCurve for drawing it in most cases now) is set to be quite low
(i.e. at 0.1 frame increments). This amounts to at most 10 sub-steps.
In one test file (with a wide window), this had the effect of improving
the performance by over 3x. It's still not as good as a sampling-free
approach, but for this functionality is still needed for FModifiers,
so it's better that we can optimise this.
Small optimisation (which shouldn't have much of an effect) where we skip
complex handle calculations if all the handles/verts for a Bezier curve
segment are all flat.
Patch by Campbell (T40372 -> F91346)
When the active action is a NLA strip, the keyframe indicator colors for buttons
and the 3D view indicator (i.e. the current frame indicator changes color) didn't
work correctly. This was because they were still checking for keyframes in
"global" time space, whereas they needed to be applying NLA corrections to
"look inside" the remapped action.
In fact, any button controlling a whole array of values were broken
because they always only keyed the index of the single fcurve returned
by `ui_but_get_fcurve()`, now pass button's rna_index value instead.
Recent flag re-order broke it since bits overlap, but logic here was far too complicated & fragile,
Checked the type of each button when testing which direction to handle events as well as block direction.
Now store the block-flipped state as a flag.
It appears it's not really needed for convenient debugging when
using proper flags passed to the compiler. Basically, it is -g3
and set breakpoint to a function as if it's not in the namespace.
Not as if a code was any wrong, just it's possible to have more
clear solution for the issue i've tried to solve in the past.
The Mesh Tools have quite few crucial tools that're missing from the toolbar. This is the main one.
The tools that're here should also be reorganized a bit to introduce actual orgnization, as it's quite sporadic at the moment. Will do that later.
Nothing related to rigify actually, recent hack in py handling of IDProp (rB3346ab03)
was breaking integrity of IDGroup's listbase of children IDProps...
Took me hours to nail this down, should have bisected for once. :/
That's like really a bummer, because currently animation data for armatures
might want to use pose, and pose might be missing on the object.
This happens when changing visible layers, which leads to situations when
pose is missing or marked for recalc, animation will change it and then
object update will restore the pose.
This could be solved by the new dependency graph, but for until then we'll
do an extra pass on the objects to ensure it's all fine.
It's done in the scene_update_for_newframe() to solve possible issues with
the render engines as well.
This finally solves issues we had with Caminandes team, where Koro would be
at the scene origin instead of being properly posed.
too crowded.
UVs in the same layer can be used for many images. It used to be
possible to filter UV faces based on the image, but this is impossible
now due to the way the system works, so I added an option to allow
filtering UVs based on active material index.
Rationale on using option and not being smart here (options are bad tm)
is that for some workflows, such as preserving image space by using the
same image for many materials, people might want to turn this off.
layer index was being obtained for loop data types but we referenced
Tessface data types
NULLing those out since only the data offsets are used in edit mode and
address sanitizer complains about freed memory access.
Also minor comment in texpainting
Murmur2a is a very fast hashing function generation int32 hashes.
It also features a very good distribution of generated hashes.
However, it is not endianness-agnostic, meaning it will usually generate
different hashes for a same key on big- and little-endian architectures.
Consequently, **it shall not be used to generate persistent hashes**
(never store them in .blend file e.g.).
This implementation supports incremental hashing, and is a direct
adaptation of reference implementation (in c++):
https://smhasher.googlecode.com/svn-history/r130/trunk/MurmurHash2.cpp
That cpp code was also used to generate reference values in gtests file.
Reviewers: sergey, campbellbarton
Reviewed By: campbellbarton
Projects: #bf_blender
Differential Revision: https://developer.blender.org/D892
Some brushes really do the same thing and we have agreed not to offer
extra presets for one brush type. Removed those brushes from default
.blend. They are Polish (Flatten Contrast does the same), Brush (Does
the same as draw) and Draw from texpaint (where texdraw/draw does the
same)
Create unique flag for output shaders with displacement data and use it
to calculate transformed normal. Implementation suggested by Brecht Van
Lommel.
Reviewers: brecht
Differential Revision: https://developer.blender.org/D890
Only fixes the crash actually, real issue is, vparent does not handle deletion of vertices
at all currently... We'd need either some kind of static uuid for vertices, or some
mapping helpers used each time we remove or reorder verts... ugh.
Org patch by Severin (Julian Eisel).
Make the UI API more consistent and reduce confusion with some naming.
mainly:
- API function calls
- enum values
some internal static functions have been left for now
The idea is to avoid memory allocation when only one segment step is to be allocated.
This gives some speedup which is difficult to measure on this trashcan from hell, but
it's about from 7% to 10% in the extreme case with single volume filling the whole of
the viewport. This seems to depends on the phase of the bug-o-meter in the studio.
On the linux boxes it's not that spectacular speedup, it's about 2% on my laptop and
about 3% on the studio desktop. This is likely because of the awesomeness of jemalloc.
Most of the unused functions were removed. Some of them were if-defed
because they are referenced from the code which was already if-defed.
Reviewers: lukastoenne, campbellbarton
Differential Revision: https://developer.blender.org/D868
Add compile-time check for particular glibc version which fixed the issue.
This makes it so own-compiled blender is the fastest in the world, and the
only issue remains what should we do for release builds.
After some discussion with Campbell we decided to keep it as is for now
because slowdown is not that much noticeable. We'll disable this workaround
for release builds when all the majority of the distros will switch to the
new version of glibc.
Even though GLSL allows to have polymorphic functions our codegen
is not aware of this at all.
Let's rename the functions for now, but in the future would be handy
to make codegen aware of the polymorphic functions.
Decouple color picker hsv data from the whole block. Basically, each
color picker now takes care of creating its own color picker role. For
this bug report it can be seen that probably HSV is not the best space
for gamma/lift/gain workflow because it is bounded at 1.0 but this is a
separate issue.
Looks like material node trees are stored directly in the material. The
reason I thought this was fixed was because my test file didn't connect
the lamp data node in the rest of the tree.
Thanks to Campbell for catching this :)
To ensure there's space for more than a few characters in the rename text fields of the File Browser, we now use a width relative to the column width (for all display modes).
Includes some edits from @campbellbarton - thanks for this!
in practice often we want to breakpoint here (instead of flooding output)
but its confusing the option does nothing in release builds.
Devs can comment out locally.
* The `NODE_OT_parent_clear` operator has been removed. This was a very
simplistic operator that detached every selected node, which is not very
useful in case of hierarchical frames. The `NODE_OT_detach` operator
only detaches the top parent nodes in the selection, keeping the
hierarchy of selected nodes intact.
* The `ALT+P` shortcut has been reassigned to the `NODE_OT_detach`
operator which replaces the previous `NODE_OT_clear` mapping with
similar behavior (also gives a menu entry shortcut now).
* Shortcuts for `NODE_OT_detach_translate_attach` have been removed, due
to crowded and messy keymap and unintuitive shortcut `ALT+F`. This macro
operator is still registered, in case hardcore users want to make their
own keymaps, but not mapped by default. Node keymaps may need some
redesign in the future for these things.
Quite striaghtforward implementation, with the only weird thing that for some reason
my video driver wasn't happy with calling the function "clamp" giving some weirdo
shader compilation error messages.
Called the GPU function clamp_val which can handle float and vec3.
That code was mainly needed for the transition period, now we've
got all platforms updated to new OSL.
Plus there are some crucial fixes baking in the current upstream
sources which we'll need to have for the next Blender release.
Even tho it's not 100% clear when we'll switch to OSL-1.6 we'd better
start preparing earlier for this, so we don't spend time on this later.
Plus this code helps troubleshooting some OSL issues, which requires
testing with latest versions of OSL.
It now supports different scheduling schemas: dynamic and static.
Static one is the default and it splits work into equal number of
range iterations.
Dynamic one allocates chunks of 32 iterations which then being
dynamically send to a thread which is currently idling.
This gives slightly better performance. Still some tricks are
possible to have. For example we can use some smarter static scheduling
when one thread might steal tasks from another threads when it runs
out of work to be done.
Also removed unneeded spin lock in the mesh deform evaluation,
on the first glance it seemed to be a reduction involved here but
int fact threads are just adding value to the original vertex
coordinates. No write access to the same element of vertexCos
happens from separate threads.
Ghost depends on glew-mx, so glew-mx should be passed to linker after the ghost.
We're also using spaces for indentation in python, including scons rules.
In cases when the subsurf modifier is the last in the stack and there
are only deformation modifiers before it we can skip doing full orig
vertex lookup.
This is rather common situation here in animatic.
The issue was caused by GLEW MX enabled in SCons by default so
basically previous commit already fixed the crash. But we need
to be safe here.
For now the fix is simple and not that clean, just check if
there's an OpenGL context available and if not we don't do any
GLSL magic.
This is to be cleaned up after some discussion with the viewport
project guys.
Had to fix it by hand (restting them to default values), theme maintainers,
please check your theme from time to time and track such ugly 'pitch black' fields.
And we devs should run theme update script much more often too - will do in next commit!
This mainly happens when over-saturating already saturated color.
After some discussion with Campbell and loads of tests we decided
to clamp the result RGB color. As an alternative we might want to
clamp corrected HSV values instead, but that would lead to some
larger changes in the render results.
TODO: The same is to be done for compositor nodes.
Add simple uvs now does a cube unwrap and pack operation. Result is not
optimal by far but it should not result in crashes and it will be quite
usable for simple cases.
The root of the issue comes to the way how we sample the gaussian filter in
RE_filter_value(). We need to scale x to -3*sigma..3*sigma segment in order
to get the whole bell.
The old code tried to do it, but failed dramatically, plus it used some weird
gaussian sampling formula. Replaced it with much more clear one, which gives
proper blur now.
There's no visible different in AA sampling in BI render tho.
Other filters like Mitchell still tends to give wrong square shaped blurs,
but they're much more difficult to resolve because they're just wrong in
the code -- for some reason smaller kernel size means more blur. Let's solve
this later.
Constraint space conversion ignores object scale, which is OK in most cases. But here,
we are converting a normal from world to local space, and when later converting it
into target space to actually do the BVH raycast, we use TransformSpace which
does applies objects' scaling to normals, as expected.
Best solution here is to also take object's scale into account when converting
from local to world space.
Issue here is that seam outset was being calculated wrong and as a
result we filled too many unneeded pixels. Code can be improved here by
clamping perhaps but left it as is for now. Thanks to Campbell for the
help resolving this!
This was never ported to a new tracking pipeline and now it's done using
FrameAccessor::Transform routines. Quite striaghtforward, but i've changed
order of grayscale conversion in blender side with call of transform callback.
This way it's much easier to perform rescaling in libmv side.
The title actually tells it all, this commit switches Blender to use the new
autotrack API from Libmv.
From the user point of view it means that prediction model is now used when
tracking which gives really nice results.
All the other changes are not really visible for users, those are just frame
accessors, caches and so for the new API.
This starts the creating the new AutoTrack API. The new API will
make it possible for libmv to do full autotracking, including
predictive tracking and also support multiple motion models (3D
planes etc).
The first goal (not in this patch) is to convert Blender to use
the new API without adding any new functionality.
This API currently contanins:
- Frame accessor to access frames which are stored in Blender side.
- New Tracks implementation
- New Reconstruction implementation
Currently this API only tested on doing the same frame-to-frame
tracking as the old API allowed to do. But it also supports now
predictive tracking which is based on the Kalman filter.
Use BLI's `rotation_between_vecs_to_mat3` helper instead of own custom code,
both simplifies the code and fixes wrong handling when snapped normal was exactly
opposed to org one (i.e. 180° rot case).
This is basically just a wrapper class, which maps the generic call from the OSL spec to our closures.
Example usage:
shader microfacet_osl(
color Color = color(0.8),
int Distribution = 0,
normal Normal = N,
vector Tangent = normalize(dPdu),
float RoughnessU = 0.0,
float RoughnessV = 0.0,
float IOR = 1.4,
int Refract = 0,
output closure color BSDF = 0)
{
if (Distribution == 0)
BSDF = Color * microfacet("ggx", Normal, Tangent, RoughnessU, RoughnessV, IOR, Refract);
else
BSDF = Color * microfacet("beckmann", Normal, Tangent, RoughnessU, RoughnessV, IOR, Refract);
}
This is a very old bug and a few attempts have been made to fix it before,
Old code was checking worldspace axis flipping, instead of axis flipping WRT the axis-alignment of the object.
amount and crashes after Cycles render.
This is a hack to fix this, but at this point the system is hopelessly
broken anyway and no good fix other than total rewriting is possible.
Adding new object to RigidBodyWorld obgroup is not a good way to do that, since it only
takes effect (create rigid_body for new objects) when you change current frame.
Better to use rigidbody.object_add() operator here!
This reverts commit 507712db3f.
Error was quite an old compiler, which had further warnings/errors.
Old compilers can just have this defined in BLI_math
Operators that trigger UI events (but nothing else)
were using 'CANCELLED' making it impossible to tell if an invoke
function failed, or opened a menu.
This changes the Py API to use array lookup table.
Previously this could be very slow since it would loop over all elements.
Now the python script is responsible for creating the internal lookup table (as with C code).
This will break some scripts.
This adds a theme option for the embossing of UI widgets. By doing this users have much greater flexibility for creating nice themes. Previously many themes (particularly dark ones) looked quite bad due to the very obvious emboss. This made simpler, flat-style themes very challenging.
Closes T42228
Reviewed by @campbellbarton
First, you should unregister in reverse order you registered your operators, keymaps, etc.
Second, when registering keymaps you have to check keyconfigs are actually available (they are not in background mode).
The ones in extern/glew-es have been changed to NOTE instead of XXX
GHOST_ContextEGL.cpp: It really does seem that it is not possible to query the swap interval using EGL
GHOST_WidnowCocoa.h: The comment referring to Carbon is clearly out of date, so I removed it.
math_geom.c: The node about not using tmax again is correct, but the code is kept for a future maintainer who will need to know how to compute it if they modify that code.
paint_image_proj.c (2698): The question about integer truncation does not appear to have been resolved. It still seems to be an incorrectly implementation of rounding (I'd suggest using the round function instead of this hack).
Add Recalcuate Normals to the Faces menu, next to other shading options.
Differential revision: https://developer.blender.org/D841
Signed-off-by: Thomas Dinges
It's the same filter which is used by default by Blender Internal renderer
and it gives crispier edges than gaussian filter.
Default filter for Cycles is unchanged because it's unclear if new filter
gives more noise or not. After some further real production tests we can
consider making Mitchell filter default for Cycles as well.
This way it's easier to extend bitfields and see when we start running
out of free bits.
Plus added brief description of what SD_VOLUME_CUBIC flag means.
It is per-material setting which could be found under the Volume settings
in the material and world context buttons.
There could still be some code-wise improvements, like using variable-size
macro for interp3d instead of having interp3d_ex to which you can pass the
interpolation method.
This is the first step towards supporting cubic interpolation for voxel
data (such as smoke and fire). It is not epxosed to the interface at all
yet, this is to be done soon after this change.
Quite straightforward change, don't think some extra explanation is needed.
This gives about 15% speedup of the modifier evaluation on my laptop.
Reviewers: campbellbarton
Differential Revision: https://developer.blender.org/D836
This commit switches meshdeform modifier to use threads to evaluate
the vertices positions using the central task scheduler.
SO now we've got an utility function to help splitting the for loop
into tasks using BLI_task module which is pretty straightforward to
use: it gets range (which is an integer lower and higher bounds) and
the function and userdata to be invoked for each of the iterations.
The only weak point for now is the passing the data to the callback,
this isn't so trivial to improve in pure C.
Reviewers: campbellbarton
Differential Revision: https://developer.blender.org/D838
When resetting edit bones on cancel, they also have to reset connected
parent and child bone tips and heads respectively, since these can be
modified during the transform.
This is only an indirect fix, in fact: this commit adds a public API to check
the maximum number of a given layer type (`CustomData_layertype_layers_max()`),
and uses it to forbid too much layer creation in `CustomData_merge()`.
This only affects UVs/VCol data though, but merge behavior in itself is not a bug
actually, how user managed to get thousands of different UV layer names remain
rather mysterious...
Curve parent requires valid path to exist for curve. If the path is disabled in
the curve settings, displist evaluation will check the dependency graph whether
the path is needed for parenting.
The issue was that changing relations in the scene chagned need of the curve path
but nothing tagged the curve to update it's path.
For now use direct call of DAG_id_tag_update from set_parent(). In the bright
future we might detect such a need in flush automatically in the depsgraph.
Vertex parent was using original non-modified nurbs list, simply because
it didn't have something else to operate with.
Now we've got deformed by pre-tessellation modifiers nurbs in the curve
cache which might be used y the vertex parent.
Main moving logic is moved to new `BKE_keyblock_move()`, which makes it available from anywhere.
In addition, move code was reworked so that it only loops once on whole keyblocks list,
and it accepts arbitrary org and dest indices, not only neighbor ones.
Partly based on work by revzin (Grigory Revzin) in his soc-2014-shapekey GSoC branch, thanks!
The issue here is that if no matching winid is found, we destroy all
windows and their context with them. This will also delete the OpenGL
textures associated with the initial context, thus we lose the icons.
This patch makes sure a window is always kept for later so the initial
context does not get lost.
Thanks to Campbell and Ton for the discussion on the issue.
Loading XML module, registering etree namespaces... etc
on startup for everyone on chance someone may want to export
an SVG from Freestyle is unacceptable.
This shouldn't have got through the review.
also disable loading when built without freestyle.
Features:
* Both still image and animation rendering, as well as polygon
fills are supported.
* The exporter creates a new SVG layer for every Freestyle line
set. The different layers are correctly sorted.
* SVG paths use data from line styles, so the base color of a
line style becomes the color of paths, idem for dashes and
stroke thickness.
* Strokes can be split at invisible parts. This functionality is
useful when exporting for instance dashed lines or line styles
with a Blue Print shader
* The exporter can be used not only in the Parameter Editor mode,
but also from within style modules written for the Python
Scripting mode.
Acknowledgements:
The author would like to thank Francesco Fantoni and Jarno
Leppänen for their [[ https://github.com/hvfrancesco/freestylesvg | Freestyle SVG exporter ]].
Differential revision: https://developer.blender.org/D785
Author: flokkievids (Folkert de Vries)
Reviewed by: kjym3 (Tamito Kajiyama)
* Fill brush did not do proper srgb conversion for byte images (reported
by Sebastian Koenig, thanks!)
* Color sampling for palettes did not refresh the toolbar.
trash with float images.
Issue is uninitialized alpha in the clone brush + a clamping issue for
float images. There's still some 'swimming' and brightness issue here
but I have to do work for the weekly, will investigate further later.
Just do not use crazyspace correction with childof constraints in this case.
Note this is only a very partial fix (partial use of parent loc on some axes
is still broken in transform), a real fix would probably require a full rewrite
of constraints handling in transform code (a mere static correction matrix
just cannot work in all possible cases, we'd need a full dynamic correction system here).
Anyway, transform code as a whole is horrible. :/
Old behavior of shuffling the meta made it hard to use metas
in a complex edit since you couldn't be sure if exiting a meta would move it in the stack.
This is so-called GPU limitation boundary hit, told compiler to NOT include
volume bound function, otherwise some real weird things used to happen.
We actually might want to do the same for CPU, inlining everything is not
the way to get fastest code.
Noise function's significant bits are up to 31st bit. This should now
give the same visual result as before, minus the stripes.
Issue pointed out by Anthony Edlin, thanks!
This was way too verbose, heavily factorized the code.
Also made sure only changed data are applied (was not always the case,
especially for curves and lattices), and that we always use raw value
when only one element is affected (was only that way for coordinates).
Note I checked performances, they seem to be roughly the same as previously.
Afaik, a hidden BMesh element should never be selected. And this check was done in an
awfully inconsistent way! If it should really be needed, should be added back as a macro
used everywhere!
Mostly reorganizing mesh data handling to be always in the same order, this piece of code
is rather verbose, let's try to keep it organised a bit.
Also some visual UI tweaking.
The sequencer 'display' area is a region on its own, so we can't hide
the preview regions. The only problem is that the <-> resize icon shows
in the main region, so you only see it over the tracks region in the
sequencer, I'm not addressing this though.
Organize Maximize/Fullscreen mess and add a new fullscreen mode with no UI
* Maximize Editor: (old Ctrl+Up)
* Full Screen Window: (old Alt + F11)
* Full Screen Editor: new operator (Alt + F10)
* Change Show/Hide Header: (Alt + F9)
When the mode is on moving the mouse near the top right corner of the
editor shows an icon to go back to the normal editor mode.
This was originally intended for the multiview branch, but this
functionality also benefits non-stereo workflows, thus it can be
reviewed and committed independently.
Development notes:
* This includes cleanups in the code to sanitize the naming of
fullscreen/maximize across the window/editor code.
* Originally the idea was to make the window fullscreen as well, but
this idea was dropped.
* You can see the clicking area when debug is 1
* Technically the user can be left with an unfaded icon in the corner
(specially when using a tablet). If we think this is too bad we can
increase the action zone to be the whole screen, or something similar.
Reviewers: campbellbarton [1], ton [2], fsiddi [2]
[1] actual code review
[2] design review
Differential Revision: https://developer.blender.org/D678
Quite annoying, the same thing we do from the blender side, But as a positive
side we can get rid of some utf8/utf16 conversions.
Hopefully it all work fine now, at leats works on mu russki windoze laptop.
* Only use last key for pies if it hasn't been released already
* Confirm threshold is now measured as distance after regular threshold.
zero disables.
* Only display the confirm threshold if there's a valid direction (mouse
is after threshold).
* Calculate confirm threshold taking recentering into account
Make sure the 1.0 value is not affected by smoothing, and min/max limits
never go above or below 1.0 respectively. This was a request by
animators since not modifying the mesh in its rest pose is regarded as
crucial.
The nodes wire was using 'TH_HEADER' flag to get its color and thus would be in sync with the header. Now make it so it uses its 'own' flag (actually 'TH_SYNTAX_R', the only TH_SYNTAX_* which wasn't yet used by the nodes).
Also expose the setting to the user so it can be themified.
This fixes T42209
Reviewers: lukastoenne
Reviewed By: lukastoenne
Maniphest Tasks: T42209
Differential Revision: https://developer.blender.org/D827
The reason being, with the current system of quick selection, it's
possible to spawn an operator after confirmation, especially for cases
with modifier buttons.
Currently the volume variation feature in stretch constraints is
unlimited. This has to be compensated by riggers by adding scale limit
constraints, but these are unaware of the stretch orientation and can
lead to flipping. Also the stretch calculation itself is not working
properly and can lead to collapsing volume.
The patch fixes this with several modifications:
- Interpret the volume variation factor as exponent, which works better
with large values for artistic purposes.
- Add integrated limits to the volume "bulge" factor, so secondary
constraints for compensation become unnecessary
- Add a smoothness factor to make limits less visible.
Eventually a generic volume preservation constraint would be nicer,
because multiple constraints currently implement volume variation of
their own. This feature could actually work very nicely independent from
other constraint features.
Differential Revision: https://developer.blender.org/D826
The reported issue was caused by a backward incompatibility due to careless
code changes made when per-material Freestyle line colors were introduced
in rB7915d7277ac8c605f016f30f943080556244fb59. In 2.71 line style
Material color/alpha/thickness modifiers was retrieving alpha transparency
from Material.alpha, whereas in 2.72 Material.line_color[3] was referenced.
The present fix reverts the aforementioned code changes.
This issue is a regression from 2.71, so the fix is appropriate for inclusion in
the 2.72a release.
This commit adds a confirm threshold property to pie menus.
Basically, this will confirm the pie menu automatically when
the distance from the center of the pie exceeds that threshold without
a need to release the pie button.
The confirm threshold will only work if it is larger than the pie
threshold.
The confirmation actually occur when the mouse stops moving, to
allow multiple pie menus to be better linked together, (see below)
This functionality also facilitates the ability for chained pie menus by
dragging. Basically, a pie menu item can be a call_menu_pie operator and
the new pie menu will still use the original pie menu release event for
confirmation. This should allow for quick, gesture based navigation in
pie menu hierarchies (going back in the hierarchy is still not supported
though)
There will be a demonstration pie in the official add-on soon
* Consider non-manifold edges as sharp, as in split normals handling.
* Consider edges from sharp polys as sharp!!!
* Fix returned number of groups (was off-by-one for non-bitflags grouping, could also
be wrong in case of id overflow).
Note about using sharp polys too to define groups: Only current use of this function
(Obj exporter) does not need that, because it does its own check for sharp faces.
However, we might reuse that func in other places in future (e.g. in custom split
normals area), so better to get a consistent behavior!
Navigation mesh object need to access the current scene at
creation time. This can be at scene start or when an object
is instantiated from an inactive layer. The method of getting
the scene differs in these cases. This fix handles both.
Issue was, parenting with operator, then unparenting would keep the inverse parent matrix.
So if you then parented again through the mere Object field of Object buttons, you'd still
use previous inver parent matrix, giving some weird behavior from user PoV.
This commit simply makes sure inverse parent matrix is always reset to indentity when clearing
parents.
In drag style pie menus clicking is ignored now. This is done because
if we want to ensure that pie menus will not respawn, we had to not
destroy and keep the pie menu invisble while the button is pressed.
This opened a can of worms where pie items that would spawn popups could
steal the focus from the invisible pies and then the former pies would
be stuck in an invisible state, stealing all input unless user did a
right click to cancel the pie.
Invisible state is still kept for cancelling and strictly for that only.
This still allows nested layouts but only in click style, while hold
style is used for faster interaction. As a plus, it should also be
possible to tweak slider properties now, though menu will close after
that.
Also reduce number of branching and multiplications a bit by inlining the branches.
This gives an unmeasurable speedup, which is in case of BMW is about 2% here.
quit.blend.
This will use a slower file write if an object is in edit or sculpt
mode.
Autosaving will explicitly not be supported to keep it fast.
Added a tooltip warning.
A StrokeVertexIterator ignores the first element when it is the only element.
Such an iterator can be created by the .incremented() method from an iterator
over two stroke vertices.
This problem is a regression from 2.71. The present fix is appropriate to backport
if Blender 2.72a is planned.
Problem report by Kazuhiro Murakawa through personal communications, thanks!
Not as if it gives any huge benefit to end artists, but it's kinda silly
to have like 3 different task/thread scheduling systems. The new one is
what's recommended for use now i'd say.
On the other hand it has unmeasurable memory save benefit :)
This gives more precise information about memory usage which might be real handy
when doing memory optimization.
It works good here for as long as i can tell but if for some reason you'll be
experiencing some weird slowdown please let me know.
Basically, this commit changes pie menu click interaction so that
confirmation is done on left click release instead of press. This allows
dragging on the pie menu to select different items, but most
importantly, there should be no left over click events passed on to
subsequent menus/pies. This means that pie menus should now be able to
spawn popups safely. Also, left clicking to spawn a second pie menu now
sets that menu to click style by default allowing for better interaction
between hierarhies of pie menus.
Two fixes here (only the second one is strictly needed to fix the issue,
but both make the system better).
First is introduction of a random generator array for use with threaded
systems where each thread needs to access its own number generator.
The random texture now uses this so it should not be influenced by other
random generator reseedings of the main random generator like it did
before.
Second, I reshuffled the texture code to resample the upper bits of the
random number first. According to Numerical Recipes, this is where the
most variance can be found, so by sampling those we avoid correlation
issues. Also, multiplying here is not ideal because if a pair of bits
are zero, then the whole result will also be zero.
Overall this is much more random (tm) than before, however result will
also be brighter, since we now have less black spots. Tweaking the
brightness/contrast should somewhat fix that, generally having the same
result as before is not possible anyway if we are to really fix this.
Also, seems like exposing procedural depth might be nice here since it
influences the precision of the texture lookup.
Not as if it gives noticeable changes render-time, but it's just weird to
convert float4 to float 3 to just access individual x/y/z components.
Plus some compilers might be more stupid than GCC and don't optimize this
out well.
Basically the title tells it all, quite straightforward implementation.
The only thing is the image.render_slot which used to represent the active
render slot index is now moved to image.render_slots.active_index.
Reviewers: venomgfx, campbellbarton
Differential Revision: https://developer.blender.org/D821
This patch removes several stroke shaders written in C++ that are unused
and don't serve a real purpose any more. The removed shaders are:
- BPy_ColorVariationPatternShader
- BPy_StrokeTextureShader
- BPy_TextureAssignerShader
- BPy_ThicknessVariationPatternShader
- BPy_fstreamShader
- BPy_streamShader
and a few more that weren't even exposed to the Python API.
Some minor edits were made by the reviewer.
Differential Revision: https://developer.blender.org/D801
Reviewed by: kjym3
This way it is possible to define render border directly from the image editor,
which is useful for at least three things:
- If viewport is really optimized out (simplified etc) then it might be hard to
guess which exact area you're mainly interested now.
- No need to switch to the viewport to do render border tweaks, could be useful
when doing compositing.
- If one need to look at particular pixel(s) which is real handy for debugging
render engines (both Cycles and BI).
Reviewers: campbellbarton, venomgfx
Reviewed By: venomgfx
Differential Revision: https://developer.blender.org/D820
We queried the wrong value when looking for the bound 2D texture. This
is not totally robust because currently bound texture may not be a 2D
one, but this should work for now.
After discussion with cambo here we decided it's better to choose arbitrary side of the box
(in this case it's X-axis) and use image from it. That's better than doing a blackness.
P.S. This is literally a corner case anyway.
The solution is to do the multiplication with the energy in the shader
after texture application.
We might be able to avoid setting dyncol completely, but this needs
better investigation. Some shader paths also look a bit redundant.
Also, texture mapping is not supported very well for light lamps, might
also need investigation.
Do not generate materials/images/UVs if they are missing.
Now we spawn a panel ("Missing Data") with operators to generate the missing data and
pop a warning if user tries to paint without them.
The reason we have reverted this is that it is too easy to end up with more textures
than we wanted. It was impossible to enter texture paint without having textures added,
and code makes too many assumptions about what user may want.
Discussed during Sunday's meeting.
This might be a candidate for 2.72a but I'm not sure how other artists will take this
(and how refined and crash-free it is), better make a few iterations first.
And for interested parties...test please, don't wait until after a release to poke with such issues.
Also, add slot operator now adds a new unconnected image node in cycles. Only
used in the "Missing Data" panel. This should be a separate commit but I am squashing it into the same commit because
it relies too much on changes done here and can be reverted easily if complainstorm occurs again.
This is only a (hacky) partial fix, actually, since `RNA_property_animated()` will still
not work in those cases... Better that than nothing, though.
Thanks to Campbell for review.
Pretty much straightforward change which gives around 30%
speedup on my laptop and around 2x speedup on desktop in
the BI (which uses gts580). Tested with huge blurs (like
10% of blur) which was rather common during Caminandes.
For now OpenCL is only limited for blur size more than
100 pixels.
This is a bit experimental still, feedback is welcome.
Reviewers: jbakker, lukastoenne
Subscribers: ton
Differential Revision: https://developer.blender.org/D576
Ray actually should have infinite length, so we can detect camera in a volume
which is bigger that the far clipping of the camera.
This might also give some speedup (wouldn't expect much tho) because we don't
need to re-calculate ray direction and length after every bounce now.
basically we skip all non-volume objects now in the volume stack function.
Depending on the show it might give some percent of speedup.
Most of the speedup would be gained in the scenes when having SSS object
intersecting the volume and taking a reasonable amount of frame space.
Single precision exponent on 64bit linux tends to be order of magnitude slower
than double precision version even with single<->double precision conversion.
Some feedback in the mailing lists also suggests that logf() is also slow, but
this i didn't confirm here in the studio yet.
Depending on the shader setup it gives ~3% with the secret agent shot and up to
around 15% with the bmw scene here.
Quite straightforward change, the only annoying thing is that we can't use
indentation for include directive just because of the way headers inlineing
works for OpenCL.
Might do smarter job in path_source_replace_includes() but don't want to
spend time on this yet.
* On sm_30 and above there is no change (was not inlined already before), this just fixes a speed regression from yesterday. 6359c36ba4
* On sm_2x (tested with sm_21), I get a nice 8% speedup in the bmw scene with this. As a bonus, cubin compilation time and memory usage is significantly reduced. Regular cubin size went from 2.5MB to 2.0MB, Experimental one from 3.8MB to 2.5MB.
Currently only summed number of traversal steps and intersections used by the
camera ray intersection pass is implemented, but in the future we will support
more debug passes which would help checking what things makes the scene slow.
Example of such extra passes could be number of bounces, time spent on the
shader tree evaluation and so.
Implementation from the Cycles side is pretty much straightforward, could only
mention here that it's a build-time option disabled by default.
From the blender side it's implemented as a PASS_DEBUG with several subtypes
possible. This way we don't need to create an extra DNA pass type for each of
the debug passes, saving us a bits.
Reviewers: campbellbarton
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D813
Was hooked up last year for testing purposes, as we already had some code for it, but the closure itself is not really good nor really useful, so let's remove it.
This way there's much less cross-references between objects and meshes
device update functions.
The only thing remained s the object bounds calculation which is needed
by bvh update. This could also be decoupled, but it's not that crucial
yet because its's how it used to be for ages now.
This adds an AABB collision check for objects with volumes and if there's a
collision detected then the object will have SD_OBJECT_INTERSECTS_VOLUME flag.
This solves a speed regression introduced by the fix for T39823 by skipping
volume stack update in cases no volumes intersects the current SSS object.
This means it's no longer needed to enable experimental feature set in order to
have proper camera in volume support. And this also means if there's something
wrong going on, or if there's speed regression for cases when camera is obviously
not in the volume -- this issues are to be reported and handled in the regular
matter.
Happy blending!
The idea is to only count intersections with objects which has volumetric shader
and ignore all other objects.
This is probably as fast as we can go without involving some forth level magic.
Now we do much better preliminary check for panoramic camera is inside the
volume object boundings.
Also we're now cacheing the has_volume in the mesh, which makes it unneeded
iterations for each object's shaders.
Should be no functional changes, just faster sync and panoramic-in-volume
rendering.
Using matcap with a displist is not really good thing to try
and it's actually a regression since we've enabled matcaps
for all objects instead of just an active one.
The issue was caused by the fact that we never used to store the
generated image color in DNA, so image reload will loose this
information.
Now we store the color in DNA, making ti so re-loading the image
will preserve it's generated color.
It is now also possible to change generated image color using the
color swatch in image properties after the image was created.
The view map is mostly treated as a read-only data structure by line stylization
operations (i.e., selection, chaining, splitting, sorting and stroke creation). The
only exception is the chaining operation in some cases where insertion of extra
FEdge objects is necessary to ensure the continuity of underlying FEdges from
which a chain is constructed.
The present revision addresses the removal of extra FEdges so to keep the view
map clean and suitable for reuse in subsequent render frames.
New render layer option named "View map cache" is added to reuse a
previously computed view map for subsequent rendering. The cache is
automatically updated when the mesh geometry of the input 3D scene has
been changed.
This functionality offers a major performance boost for Freestyle
animation rendering when camera-space mesh geometry is static, as well
as for repeated still renders with updates of line stylization options.
Although the "View map cache" toggle is a render layer option, the cache
memory is shared by all render layers and scenes. This means that if
Freestyle is used for two or more render layers (possibly in different
scenes through the compositor), then the cached view map for one render
layer is replaced by a new view map for another render layer and hence
no performance gain is expected.
We on'don't really support 10.5 anymore for my knowledge and in any case
the defined symbol was never checked in the sources, so this workaround
was basically doing nothing for quite some time now.
This makes headers a bit less clean (with the anonymous structs
mainly and headers inclusion from another header.
Makes compilation with clang a bit less noisy.
The following two sort keys are added for sorting chains.
* Projected X - Sort by the projected X value in the image coordinate system.
* Projected Y - Sort by the projected Y value in the image coordinate system.
A new line style option for the selection of first N chains is also added.
Moreover, the chain sorting and chain selection operations are now executed
in this order instead of the reverse order used previously. The UI has also
changed accordingly. This functional change is backward compatible and
won't result in visual differences.
Some nodes only work in certain node trees, so don't show them in the Add Node menu when this is the case.
This can probably be expanded to Input Nodes too, but need to double check some cases here still.
On 4k devices the default pixel size leads to tiny OpenGL drawing
that is hardly usable without doubling the DPI. The retina system
on OSX aims to alleviate this problem by introducing a general 2x
pixel size.
No equivalent feature exists on other platforms so far. However,
to emulate the effect this patch introduces a "virtual" pixel size
factor for OpenGL drawing.
Note that the user currently has to enable this manually by selecting
the "Virtual Pixel Mode" in the user preferences (defaults to native).
All windows of a Blender instance share the same virtual pixel size as well.
It may be possible to handle this on a per-window basis and automate
the selection somewhat (if enabled by the user), so working with
multiple screens becomes more convenient, but technical limitations
make this a bit difficult (on X11 with nvidia drivers the actual screen size
is not reported correctly).
Reviewers: ton, campbellbarton
Differential Revision: https://developer.blender.org/D669
Collection of test cases from libGDX and our own tracker
Tests:
- combine triangle area matches polygon area.
- tris have same winding.
- tris don't have duplicates.
- correct number of internal & boundary edges.
- degenerate polys still give topologically correct output.
also checks all possible start-vert offsets, forwards and backwards.
optional OBJ output, for debugging.
The Operators.reset function is exposed to the Freestyle Python API, which makes
it possible to combine multiple style modules into one file.
Differential revision: https://developer.blender.org/D802
Author: flokkievids (Folkert de Vries)
Reviewed by: kjym3 (Tamito Kajiyama)
Update lzo library due to security issue
Update lzo to latest version. This fixes a security issue (reported in T41974)
This also fixes some (false-positive) valgrind warnings.
Reviewers: sergey
Differential Revision: https://developer.blender.org/D803
Select operator that takes multiple selected face regions and
selects any number of matching regions (when they have distinguishing features to isolate them).
UI access next.
Issue was caused by the launcher not dealing with slashes in the way
windows expects them to be handled -- last slash of the path considered
an escape character for the following qoute.
This is definitely to be ported to the 2.72 release.
This helps to improve the accuracy of UV unwrapping and laplacian deform for
high poly meshes, which could get warped quite badly. It's not much slower,
doubles are pretty fast on modern CPUs, but it does double memory usage. This
seems acceptable as otherwise high poly meshes would not work correctly anyway.
Fixes T39004.
It works around strange shading bug when building with MSVC.
If such weirdeness continues, we perhaps would need to use
proper inline flags all the time.
Anyway, lets see how things will behave now.
Instead of having a label which basically duplicated the information
about experimental feature set being used (which had a bug because
it claimed experimental GPU kernel is used even if compute device is
CPU btw) now we've got an enum item icon.
So once you switched to experimental feature set you'll see an
exclamation mark icon in the enum, so you know something might be
unstable or slow.
Updating maximum requires a bit of a cycle which usually does 1 iteration only,
sometimes needs a bit more but seems there's no speed regressions.
For now the code is commented out. This way it's easier for others to verify
there's no speed regressions.
Reviewers: campbellbarton
Differential Revision: https://developer.blender.org/D626
In order to compile the new kernel you need to specify sm_52 in SCons / CMake, and use CUDA Toolkit 6.5.19, from here: https://developer.nvidia.com/cuda-downloads-geforce-gtx9xx
Note: sm_52 is not enabled per default yet, so it won't be bundled with the Buildbot builds. That will be addressed later.
Basically the title says it all, volume stack initialization now is aware that
camera might be inside of the volume. This gives quite noticeable render time
regressions in cases camera is in the volume (didn't measure them yet) because
this requires quite a few of ray-casting per camera ray in order to check which
objects we're inside. Not quite sure if this might be optimized.
But the good thing is that we can do quite a good job on detecting whether
camera is outside of any of the volumes and in this case there should be no
time penalty at all (apart from some extra checks during the sync state).
For now we're only doing rather simple AABB checks between the viewplane and
volume objects. This could give some false-positives, but this should be good
starting point.
Need to mention panoramic cameras here, for them it's only check for whether
there are volumes in the scene, which would lead to speed regressions even if
the camera is outside of the volumes. Would need to figure out proper check
for such cameras.
There are still quite a few of TODOs in the code, but the patch is good enough
to start playing around with it checking whether there are some obvious mistakes
somewhere.
Currently the feature is only available in the Experimental feature sey, need
to solve some of the TODOs and look into making things faster before considering
the feature is ready for the official feature set. This would still likely
happen in current release cycle.
Reviewers: brecht, juicyfruit, dingto
Differential Revision: https://developer.blender.org/D794
Basically the title says it all, we need to update volume stack when doing ray
scatter for SSS. This leads to speed regressions in cases scene does have both
volume and SSS (performance in case there's no SSS or no volume should be the
same).
We might try optimizing kernel_path_subsurface_update_volume_stack() a bit by
either recording all intersections or using some more appropriate visibility
flags.
Reviewers: brecht, juicyfruit, dingto
Differential Revision: https://developer.blender.org/D795
Ghash comp callbacks must return false in case a & b are equal!
Also slightly cleaned up gash code using those comp func,
since those return booleans now, let's compare tham against booleans!
This is rather useful to see how good optimization went and so.
Currently uses quite simple notation: shader nodes are nodes on the
graph, connects between graph nodes are named by the sockets names,
so i.e. connection between BSDF and Mix would be named bsdf:closure1.
Could be improved in the feature to draw fancier graph, but it's good
enough already.
Use in the following way:
- To create graphix file call graph->dump_graph("graph.dot")
- To visualize the grapf call: dot -Tpng graph.dot -o graph.png
This commit makes it possible to use Glog library for the debug logging.
For now only possible when using CMake and in order to use the logging
the WITH_CYCLES_LOGGING configuration variable is to be enabled.
When this option is not enabled or when using Scons there's no difference
in Cycles behavior at all, when using logging and no output to the console
impact is gonna to be minimal.
This is done in order to make it possible to have debug logging persistent
in code (without need to add it when troubleshooting some bug and removing
it afterwards).
For now actual logging is not placed yet, only all the functions needed for
the logging are written and so.
This would give some performance boost when solving huge scenes,
amount of boost depends on particular scene.
For now enable all of the specializations, in the future we might
add some local patches and restrict only unneeded ones.
Attempt to select closest bones when possible.
Occlusion query selection does't support this well because we can't
really derive depth information from occlusion tests. May be possible to
improve this somewhat in the future.
tri_shader does no longer need to a float.
Reviewers: dingto, sergey
Reviewed By: dingto, sergey
Subscribers: dingto
Projects: #cycles
Differential Revision: https://developer.blender.org/D789
Basically the same as AC2c58e96685e8, but for Mix RGB Shaders, in case we use the Mix type. This way the node can be used as texture switch for example, setting the Factor to 0.0 or 1.0, without wasting extra memory / render time.
The UV values includes the image width/height now. To restore the
previous method as close as possible (even though it is not documented
anywhere how this is supposed to work), we have to ignore this scaling.
This ensures that the beams color does not darken along borders,
by using the last valid color of the ray as the border color (extending
colors in the direction of the source point).
- Ortho result from view3d_utils.region_2d_to_vector_3d was flipped.
- Persp result wasn't normalized.
- operator_modal_view3d_raycast.py failed for ortho views.
Thanks to Philipp Oeser for the initial fix.
This is not needed when we tag normals as dirty, but in simple cases, if offset has some
rotation (or non-uniform scaling) component, vertex normals has to be corrected.
Not really sure what the issue here was initially (might have been a
driver problem) but looks like the issue is solved if we skip generating
a display list for clipped objects; Driver does not like a generated
list that is not drawn.
Now we try to clip the object always, not only when using display lists.
This means that display lists will only get generated if object is
actually visible.
Also used glPush/glPop for instance matrices since glLoad can cause
recalculation of inverses in the driver.
Fixed dead references of API identifiers (e.g., freestyle.types.Interface0D)
due to relocations of the identifiers into submodules. Also made various minor
revisions of mark-ups and typos.
Incorrectly accessing the menu pointer, set it to NULL before doing the
tests.
We don't restore the menu pointer because it is invalid by the time the operator ends.
Issue, after a lot of blood sweat and tears, was found in
ui_but_update_from_old_block, where we restore a button to its old
values when possible. The problem here is that a1 and a2 are not really
meant to store temporary variables, because they tend to get overriden
and palette selection is one of those temporary states.
Instead, we now store the position of each button in the palette in a2
and pointer to the palette in the customdata pointer of each button and
use that to test if it's active. The positions won't change when
clicking so we are guaranteed that the old button won't override the new
one with garbage.
It's still hacky but it is better than testing button types when copying
old values.
Freestyle sections of the API docs were empty due to Freestyle module reorganization
in commit rB6498b96ce7081db039354228213d72e8c70bd3aa.
Module __all__ property was added to submodules so as to properly exclude irrelevant
documentation elements such as mathutils.Vector.
Currently, this pointer prop is not editable, and making it so does not seem
to be trivial (sound strip seems to use the filepath path, not the soundID pointer?),
so just hide the matching UI item.
Was using first vertex for depth (which could be anywhere),
now use the view-pivot.
was also copying uninitialized vector into knife_find_closest_* return values (confused debugging).
Added a special notifier now NC_WM|ND_UNDO in order to deal with such cases
and now compositor/image will refresh when undo happens.
There are much more ways to fail compo to update the resul, like undoing
while it's not visible and so, but as mont29 said -- let's at least fix
obvious crap in the workflow.
- only include files known to git.
- includes all submodules.
- version extracted from BKE_blender.h for naming.
- MD5 checksum generated.
- 'make tbz' convenience target.
Script by Dan McGrath with own minor edits.
Fix alpha under equation according to porter-duff and user docs that
state that effect should invert he order of channels and reduce to alpha
over when factor is 1.0.
BTW, we don't really expose any factor here...lost feature from 2.49?
@campbellbarton, it's really helpful to have strict flags enabled
for release builds as well -- especially when you're working in areas
which are heavily using least squares solvers.
For now made it so blender compiles, but it could be better solution here.
- selecting a boundary edge would randomly point in/outside the face (now point away).
- selecting 3 verts would use the first selected edge as the tangent (now use longest).
- selecting 1 vert betweem edges, uses the edges to define the tangent.
Freestyle was using the default Cycles rendering settings (e.g., the number
of samples) and users could not change them. Now all render parameters
(except for film_transparent) are inherited for Freestyle stroke rendering.
Problem report by Danny Grimm on Facebook, thanks!
- clamp selection on entering editmode (Python may have changed).
- BKE_vfont_select_get no longer succeeds with an empty selection
this caused many operators to go ahead and operate on zero size array.
First one, reported by Sergey, was searchbox not closing when validating it (enter or LMB click)
without any active item (we need to enable escapecancel here, to close menu too).
Second one was SEARCH_MENU_UNLINK variant not unlinking at validation when no active item.
Thanks a bunch to Campbell for help here!
Tool-tips and header-text used a different pixel alignment from the rest of the UI.
This causes blurry text with multi-sample with NVidia.
Other text still needs the issue resolved (View3d info for eg)
Currently there are inconsistencies with pixel alignment.
but this commit has no functional changes.
- wmOrtho2_region_ui for UI/Text.
- wmOrtho2_region_pixelspace for 2D drawing.
- wmOrtho2_pixelspace - when the region isn't used.
- follow references to Bone, Sequence, Node (non ID types) by name only.
was causing infinite recursion.
- add support to skip individual class properties.
This can now write an entire blend-file.
Core of the issue is that pointcache handling in depsgraph were (re-) tagging
for update some objects on hidden layers, when all their dependencies remained
untag.
Since we do not want to update objects on hidden layers, take this data into account
when flushing pointcache.
Investigations and org patch by self, reviews, advices and final patch by sergey, many thanks! :)
This adds a fresnel conductive OSL preset to the Text Editor. Based on a patch by Lukas Stockner.
Differential revision: https://developer.blender.org/D145
See the differential for details.
* Improved Cycles fire, patch by Gottfried Hofmann. Differential revision: https://developer.blender.org/D777
* Added Smoke + Fire type to the operator too.
* Cleanup.
message in terminal was: RNA_boolean_get: WM_OT_append.relative_path not found.
Added check for existence of the relative path property which was removed from append
since it is not needed and used.
* Add Density multiplier.
* Use Smoke color attribute.
Patch by Gottfried Hofmann, with some small tweaks by myself.
Differential revision: https://developer.blender.org/D776
Quick fix, this is actually a demonstration of why we should use modal keymaps!
We can give any event to *start* the op, but then hard-code how to *end* it... tsk.
* Fix caustic properties, was not updated.
* Remove wrong items, leftovers from panel splitting.
* Add missing items. Even if the bundled presets do not set those, a user expects that all properties inside the panel are taken into account, when adding a new preset.
Node was simply ignored by occ shading (noted as TODO), though it's a mere matter
of a very few lines of code, nowadays... Just copied from similar task in bake code.
Those two mimic our BLI invert_m4_m4_safe - they add a small offset to diagonal values,
in case org matrix is degenerated, and if still non-invertible, return identity matrix.
Org patch by me, final enhanced version by ideasman42, many thanks!
Move projection paint data ensure function to projective texture
painting. Also, cleanup draw data when generating a new image for
painting in image paint mode.
Since the choice to link or append has been removed in the file browser operator panel,
there was no way to tell whether as a user you were linking or appending.
To fix this the proposed patch separates the operators.
Reviewers: campbellbarton, carter2422, venomgfx
Subscribers: fsiddi
Maniphest Tasks: T41593
Differential Revision: https://developer.blender.org/D770
This was never implemented, actually... Was also the occasion to rework a bit
`curve_deform_verts()`, code there was sometimes needlessly complicated and slowish
(like getting weight from dm vertex per vertex, instead of whole array at once...).
In this case, we want to use defaults as defined by UI code (among other, from order of registering),
there is no point in storing those data in factory startup.
This reverts rB52c06440d8e51c8661a679bcb33742666ce8dbf9 and rBe40d8258bb46926a1aecf51236822532397993f3
(tabname is old 2.4x tabed panels system, *not* new 2.7x tool tabs!).
Also disabled (#idef'ed) everything regarding tabname/tabed panels for now, we may even remove it completely,
would make things clearer imho.
Note files saved with 2.71.6 up till now would have two versions of some panels in store (with two different
values for their tabname), can give some order oddities in those cases, which have to be fixed by hand...
This is rather legit case which happens i.e. when having persistent images enabled
and session is updating the lookup tables.
Now device_memory keeps track of amount of memory being allocated on the device,
which makes freeing using the proper allocated size, not the CPU side buffer
size.
Free draw objects instead of tagging object update and doing
recalculation of derived meshes. The reason is that if user deletes all
slots and tries to paint, the system will invalidate the cached in
projection painting derivedmesh. This will promptly crash. Invalidating
the draw objects only is also much cheaper.
Bring back shading in texture painting.
This works now but it uses 3 texture units instead of two. Most GPUs of
DirectX 8 (OpenGL 1.4 should cover that) functionality even should have
those, but some old GPUs might not work with that. In any case, I hope
we will be moving to OpenGL 2.1 requirement soon anyway where 4-8
texture units are usually the norm.
The problem addressed here is that there was no mean to check if an iterator
points the last of the elements being iterated over. Such checking is necessary
to reliably dereference the iterator (i.e., calling the operator*() method of the
underlying C++ iterator object).
Now Interface0DIterator and StrokeVertexIterator have an .at_last property
to check if an iterator points the last element. Using this new API feature,
the present commit partly reverts the previous commit rBeb8964fb7f19 to
better address T41464.
Differential revision: https://developer.blender.org/D752
Author: flokkievids (Folkert de Vries)
Reviewed by: kjym3 (Tamito Kajiyama)
- Drawing grease pencil fail without scene-lock.
- Converting to curve failed without scene-lock.
- Outliner drag into viewport failed with local-view.
Few things:
- reflect() takes arguments in this order: N, I, it was swapped
in the previous code for some reason.
- Normal and view vectors are to be normalized. For the view
vector we're now using shade_view() in order to deal with the
ortho camera. However, Cycles does not support ortho camera
for reflection, but this is easy to do in a separate commit.
- Reflection vector is to be in the world space. Kudos to
Antony Riakiotakis for figuring this out!
- ignore hidden faces & verts
- when cutting a pair, select edges co-linear to the cut.
Also support creating a buffer from hidden elem's even if BMO_FLAG_RESPECT_HIDE is enabled.
(if the hflag used includes BM_ELEM_HIDDEN).
as keyboard ones!!!
Note: better solution would be to move those 'internal' events to the 0x5xxx area,
but need to talk with the team to be sure this won't break something first.
Before this Cycles used to try using the cache even so it knew for the
fact that reading it from the disk failed. This change doesn't make it
more stable if someone will try to trick Cycles and give malformed data
but it solves general cases when Blender crashed during the cache write
and will preserve rendering from crashing when trying to use that partial
cache.
Issue here is that indices of edges will be incorrect when index of
vertices forming the edge is 3-0 or 2-0.
There are still issues here at uv edges where seam polygons intersect
each other but to solve that we need a better pass that detects uv edges
and pushes polygons along the "normal" of the edge instead of scaling
the polygon itself.
I don't see a reason not to do this, and this also fixes update problems when 3D View rendering is running (no volume shader), and then a volume shader gets added.
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.
There were several issues involved into triangle ribbons hair:
- Even for the viewport rendering the blender scene camera was
used for orientation. This made hair triangles oriented to
the scene camera, not to the viewport camera.
- Triangle orientation was actually supposing the camera is
perspective. Triangles weren't oriented properly for the
orthographic camera resulting in different hair width across
it's length.
This issues are solved now, but there are some related TODOs:
- Rotating viewport doesn't re-orient the triangles, so after
viewport navigation hair might not look correct. However,
with this fix toggling viewport render (to force hair sync)
makes viewport render correct.
This isn't so much trivial fix, would require making BVH
aware of the dynamic triangle orientation, so they get
properly oriented without full hair re-sync.
- Panorama camera behavior didn't change but looks like it
should, however not really sure atm what's the right thing
to do here.
Include explicit control for texturing:
This commit introduces a painting mode option, available in
the slots panel. The default value "Material" will create slots from the
blender material, same as just merged from the paint branch.
The new option "Image", will use an explicit image field that artists can use
to select the image to paint on. This will should allow painting regardless
of the renderer used or for use in modifiers.
The issue was caused by the changed defaults from the Cycles side.
Because of those properties being saved as an IDProp and not being
saved to the file, every change to the defaults would ruin someone's
day updating the values.
Added a bpy.app.handler.version_update which is run after the regular
do_versions() are done and could be sued by the scripts to apply
versioning code on their settings.
Reviewers: campbellbarton
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D761
The issue was caused by the world localization setting the ID_WORLD
type update tag. Now using a function which doesn't pollute the library
and which doesn't set update flags.
Issue was caused by the precision issues which made sdivm by 1 under
it's actual value. We can try to do some eps magic, but from the tests
on laptop and desktop doing integer division is not slower than using
floats here.
Added an extra option to `insert_bezt_fcurve()`, to allow full override of existing
keyframes when pasting (in this case, we do not want to inherit handles from existing
curve!).
This reverts commit 7c7cb01aa5.
The smoothing weights will cause shifting in the image if not calculated
as they were in the branch. Solution for better performance will be to force-clip
kernel to a square of size two for projective painting.
Thanks for Aldo Zang for the help with the fix for the panorama/fisheye
depth of field calculation and the overall math.
Reviewed By: sergey, dingto
Subscribers: juicyfruit, gregzaal, #cycles, dingto, matray
Differential Revision: https://developer.blender.org/D753
Also move this to a generic place in SConstruct instead of having this
defines in 6 different configurations.
Should not be functional changes, but please verify all the platforms.
Now we build 2 .cubins per architecture (e.g. kernel_sm_21.cubin, kernel_experimental_sm_21.cubin).
The experimental kernel can be used by switching to the Experimental Feature Set: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Experimental_Features
This enables Subsurface Scattering and Correlated Multi Jitter Sampling on GPU, while keeping the stability and performance of the regular kernel.
Differential Revision: https://developer.blender.org/D762
Patch by Sergey and myself.
Developer / Builder Note:
CUDA Toolkit 6.5 is highly recommended for this, also note that building the experimental kernel requires a lot of system memory (~7-8GB).
Remotely based on patch by kevindietrich (Kévin Dietrich), but using
a single generic panel here, as suggested by UI team.
Note we add this panel in all modes (only one tweak in scuplt mode,
where there is no history menu generated it seems, unlike other
'paint-like' modes), we can decide to move it into its own tab later.
Differential Revision: https://developer.blender.org/D733
Just have a default 'Misc' category (harmless for panels in non-cat context).
In case we would still want a panel to show in all tabs (rather unlikely),
just explicitely give an empty string to its bl_category property.
Note I choose 'Misc' because it's much shorter than 'Uncategorized' (space
is an issue here), it's a one-liner to change it anyway if UI Mafia does not like it!
whenever enter edge bevel weight editing.
This is what happens when one makes an edge sharp, for instance.
The edge bevel weight display is not on by default for performance
reasons, but seems reasonable to enable it when user indicates
interest by editing bevel weight values.
Turned out there were several issues in handling of scale parameter by numinput.
Fixed that by factorizing more some code in common with 'usual' numbuttons eval code
(new `bUnit_getScaleUnit()` helper will return valid scaled value, depending on
given system and type).
Now, numinput behaves as expected - using default unit amended by scale in case no unit is given
(i.e. entering '20' with a scale of 0.01 will give you 20cm, and '20cm' as well!).
It can be helpful in some cases and it works properly, so no need to hide it behind the experimental flag anymore. It's only enabled for the CPU though.
Also stumbled uppon 'move marker' code, was needing a bunch of fixes, cleanup and simplification,
and added a candy feature - now you will enter numinput values in seconds when editor is
in 'time' mode, instead of frames!
Turned out to be a clean/fix up of modal bevel tool, percentage mode handling was broken,
numinput handling was broken, etc.
Also added a way to switch between bevel types (modes) with M key, and tweaked a bit
numinput code to return early in case of 'char' event with ctrl modifier.
Limitations:
* Smoke/Fire rendering is *not* supported on GPU yet, that is also documented here: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Materials/Volume
* Decoupled Ray Marching is also not supported yet, so no Equi-Angular and MIS sampling yet.
Note for Builders and Developers:
* Make sure to use the CUDA Toolkit 6.5 from now on. 6.0 might still work, but can cause slower renders.
Also, tweaked versionning for lnors color - we can add immediately versionning code,
using future next subversion, even without actually switching to this subversion now.
Avoids 'empty' versionning blocks floating around, and often forgotten when actually
raising version numbers!
This was originally caused by a6ae12a where i didn't foresee unclear distinguishing
between empty and non-synced meshes will give issues for the viewport. They're the
same for final rendering, but for viewport we need to be accurate here.
This is more like a workaround actually, we use a fixed 'margin' for height in case of search menus,
instead of using shadow width (which gave the bug with low values, and insane margins with big ones).
Note root of the issue is that if 'top' margin is too small, the first entry of the search menu
gets activated before the 'opening' click is released. This means that button will get the
KM_RELEASE event, and immediately quit (see interface_handlers.c:7945, ui_handle_menu_button()).
The sunbeams node was clamping the range of influence to start at 1
pixel distance from the source. This was a poor fix for artifacts caused
by an off set in buffer coordinates. Since the u coordinate starts at
ceil(umax) the v coordinate also has to use ceil. This also fixes some
discontinuities that became visible when the source point is close to
a sharp line in the input image.
Scene replacement with invalid scene name was crashing blender,
now it's a no-op.
KS_Scene.replace() to return a boolean to indicate if the scene
is valid and is scheduled for replacement. This allows more
robust game management.
That was only needed in the beginning, when we did not had support for tangents. It's time to clean some of the defines up, it's getting a bit too much.
* __VOLUME__ is basic volume support with Emission and Absorption.
* __VOLUME_SCATTER__ enables volume Scattering support.
* __VOLUME_DECOUPLED__ enables Decoupled Ray Marching.
- Get rid of the obsolete operator
- Use select_or_deselect_all in 3ds keymap, which performs better
than separate deselect all and select binded to the select mouse.
This was already done for maya, and seems was accidentally reverted
by 5a91db3d.
Now always check for a default unit, and evaluate the whole expression in this "unit space".
Not an ideal solution, but should handle most cases nicely
(we can't address all possible corner cases anyway).
Note default unit is searched in current string first (bigger unit of current system wins),
then in previous string.
Note this also replaces ',' by '+' in default separation between units,
helps solving issues with parenthesis (e.g. (1'1")*2.5 would fail in existing code)!
This would break if someone uses py ops with lower precedence than '+' (like bitwise
operations, and comparison), but these are not expected usecase here anyway.
Reviewers: campbellbarton
Differential Revision: https://developer.blender.org/D340
The reported issue was caused by an old bug combined with another bug
introduced by recent Freestyle Python API updates.
The old bug was that a mutable reference to CurvePoint was treated as if
it were immutable. Iteration over CurvePoint objects is implemented by
the C++ CurvePointIterator class, whose dereference method
CurvePointIterator::operator*() returns a reference to a mutable data
member (probably originally intended for better performance). Hence the
returned reference may vary upon iteration over different CurvePoints.
This implementation detail was overlooked and the returned reference was
treated as immutable (which is the case in fact for other Interface0D
subclasses except for CurvePoint). This bug was surprisingly old as it
existed before the beginning of Freestyle integration into Blender.
The other bug was in the MaterialBoundaryUP0D predicate class that was
not properly handling the end of iteration. It is noted that when the
iter() and next() built-in functions are applied to Interface0DIterator,
it is no longer possible to reliably check the end of iteration by the
.is_end property of the iterator. Namely, the .is_end property works as
expected only when iteration is carried out in combination with the
conventional .increment() and .decrement() methods of the iterator. For
this reason the commit rBb408d8af31c9 was partly reverted to recover the
previous definition of MaterialBoundaryUP0D.
This dragging zone is visualized as the circle (the same as object origin)
in the spline bounding box center and dragging that circle drags the whole
spline.
Pretty much basic functionality, but quite useful in practice.
Requested by our roto team (Sebastian and Sean :) in IRC.
`WM_keymap_guess_opname()` was missing a bunch of op 'types'/familly. Now all are there,
either trying to find a matching keymap, or explicitely listed in a comment as skipped for now.
Note matching might not be perfect in all case, but we can easily tweak that later if needed.
This problem was introduced in 983cbafd18
Basically the issue is that we were not getting a unique index in the
baking routine for the RNG (random number generator).
Reviewers: sergey
Differential Revision: https://developer.blender.org/D749
The title says it all, move the EWA filter to BLI (currently it's
math_interp.c) and use the function from both BI renderer and the
compositor.
This makes more central place of the algorithm, allowing to have
fixes and optimizaitons synchronized across the two usages.
This also fixes T41440: Displacement in compositing creates holes
Reviewers: campbellbarton, lukastoenne
Reviewed By: lukastoenne
Maniphest Tasks: T41440
Differential Revision: https://developer.blender.org/D748
Root of the issue goes to the fact that bevel list calculation might drop some points
if they're at the same position. This made spline length calculation goes wrong.
Now the length of the bevel segments is stored in the bevel list, so values are
always reliable.
Initial patch by Lukas Treyer with some tweaks from me.
Modeling tool to cut intersections into geometry (like boolean, without calculating inside/outside).
Faces are split along intersections, leaving new edges selected.
Access from Face menu.
Commits early in this year (to save some space) broke this. Hopefully this time
it works in all cases - lastCageIndex is no more influenced by realtime/edit active states.
Also, inactivate buttons instead of hiding them, can be useful to set those data
even though it does not have any immediate effect.
Took the opportunity to switch cage buttons to RNA, btw.
Operators' poll func might be called from anywhere in Blender, so they should
not make any assumption about available context. material, lamp and world
are specific to context from Properties space...
For some reason, labels of buttons in popups/menus were clipped left.
Removed that function (was the only place were it was used), clipping text
by its left is really bad for its understanding!
Instead, extended clip_middle code so that it optionnally preserves a right piece of
the given string (like the shortcut part of a menu label). Think that's what makes
more sense here (note that most menus adapt themselves to their longest item,
we do not have much fixed-width ones, anyway).
Also tweaked a bit clip_middle logic, so that left part always have at least ten chars
(in addition to already existing 'min width' checks).
There was some fuzzyness in `region_overlap_fix()`, using an 'other side' region
as ref to move current one in case their rect would intersect...
New code is a bit more complex, but should handle nicely all situations, mostly
ensuring we only translate an overlap if we find a previous one **on the same side**,
and ensuring we also never have intersecting overlapping regions from different sides
(since this does not work nice at all).
It now uses the tile size to split the job. For CPU this may add
overhead, but for GPU this is highly needed.
Reviewers: sergey
Differential Revision: https://developer.blender.org/D690
Indeed, constructive modifiers are highly likely to modify CDLayers like vgroups or vcols!
See vertexgroup2.blend file ({F93770}) in T40523, especially obvious with subsurf modifier.
This solves the crash, but indices might be wrong after the separation.
There are loads of other ways to confuse hook indices in both meshes and
curves, for now considering this a TODO.
* Don't compute expf() for every step, instead sum the intermediate values and calculate it every N (8 for now) steps. This helps a few percent (~5% on a cube with wave texture) in my tests here.
If user drags away from initial position, menu changes to drag style and
returning to that position won't remake the menu click-style. Allows to
use the threshold indicator to cancel the pie.
Add precision mode to radial operator by keeping the shift key pressed.
Precision mode works by checking difference between absolute window
coordinates and the point where shift was pressed and adding those to
the distance between that point and the radial center. This allows
bigger negative/positive range than using a strict radial scheme.
Missed some passes in the previous commit. Now seems all the passes
are covered, at least with my understanding of the things.
There're some weird things going around with the refraction pass,
but that is caused by some other issue in the code. Would rather
call it a TODO for now.
Make pie menu item placement touch the radius from the internal side of
the buttons rather than placing on the center on the cirtcle. This
allows us to get rid of the separate visual angle property, also allows
for tighter placement of pies with a smaller radius without easily
overlapping.
Also pie menu title now always appears above the threshold indicator.
Root of the issue goes back to the on-fly normals commit and the
latest fix for it wasn't actually correct. I've mixed two fixes
in there.
So the idea here goes back to storing negative scaled object flag
and flip runtime-calculated normal if this flag is set, which is
pretty much the same as the original fix for the issue from me.
The issue with motion blur wasn't caused by the rumtime normals
patch and it had issues before, because it already did runtime
normals calculation. Now made it so motion triangles takes the
negative scale flag into account.
This actually makes code more clean imo and avoids rather confusing
flipping code in mesh.cpp.
This is rather workaround solution for now, which seems to
work and it's not that huge to maintain (one liner apart from
the comment).
Idea is to make sure PeekMessage peeks the message when window
proc receives WM_MOUSEWHEEL (some touchpad drivers seems to
swallow the messages making it so PeekMessage doesn't get
anything).
This is just another issue caused by convertblender overwriting the object
matrix at the time of creating render object. What's even worse here is that
original matrix is not stored for the lamps, only lamp_matrix*view_matrix is
stored.
For sure we can combine lar->co and lar->mat back to mat4, multiply by the
inverse view matrix and get object matrix, but this is not suitable for the
viewport render because every viewport rotation will accumulate the error.
For now let's store worldspace lamp matrix in the LampRen structure and use
it when rotating the scene.
Attempt to make soften brush faster by allowing non-symmetric kernels.
Projective painting supports those naturally but for 2D painting there's
a small hack to avoid shifting of the texture. Not totally correct but
it works for now.
Sampling still samples the texture color in transparent areas. This is
not so bad but users may get confused when clicking on a white spot and
picking black instead of the mesh color.
Krita also has this uncertainty when picking in transparent areas but we
do not interpolate with an explicit "transparent" looking texture during
viewport drawing (maybe we should), so it's not so apparent what happens
here.
Since 3DViews use IDs like images or clips, we can't skip anymore `lib_link_screen()`
when reading from mem for undo/redo stuff. Else, freeing (unused) screen in `BKE_read_file_from_memfile()`
will lead to using data already freed (since pointers have not been updated when reading that undo step).
There were a few issues to fix here:
* We did not really unpremultiply float image dabs prior to sending them
to the GPU. That made float and byte image result different in texture
painting and undoing could change the result.
* To make textures nicely composited over the mesh, I used decal mode in
OpenGL texture environment for the texture unit. This uses the texture's
alpha channel with a nice over operator.
* Texture creation used to override the alpha setting due to the display
restrictions. Not so anymore, people can now create transparent byte
images.
Also, made alpha zero default for new textures now, since it has such a
nice effect here.
Problem is that setup of stack indices which refer to the same stack
entry can lead to cyclic TexDelegate node pointers, causing an infinite
loop.
Fixing this would take too much time and require recoding large parts of
the texnodes system, which is earmarked for scrapping anyway ... So for
now just disabled muting in texnodes to avoid crashes.
The stored context object was used for creation of shade nodes. A closer look at the
node system showed that the context is not actually used when shader nodes are
added to a shader node tree. Relying on this fact, now a NULL pointer is passed to
nodeAddStaticNode() instead of the stored bContext pointer.
Looks like the issue was caused by a UV map name starting with a lower case
letter (e.g., "color"). Capitalizing the name fixed the problem. Also adjusted
the creation of custom data layers to optimize things a bit.
Removed the previous changes for passing a line style through the Controller, and
revised the BlenderTextureShader to assign the shader node tree of a line style
(if specified) to strokes. This way the assignment of shading nodes can be done
through both the Freestyle GUI and Python scripting.
Changes to ED_node_shader_default() were reverted since the code there was actually
not suitable for setting up the default line style node tree properly.
Now the shader node tree of a line style ID datablock is used to define textures
as well as their mapping and influence.
TODO: Textures alpha channel mapping and influence.
TODO: Blend mode in the Output Line Style shader node.
We can't really make CPU and GPU results look the same in all possible
circumstances, but here we can make them look close enough to each other
by making it so sobol pattern for bounce number is the smae for both
CPU and GPU.
This makes CPU and GPU render results look the same with low number of
samples, high number of samples was never an issue.
The issue is that we've got hell with DEBUG and _DEBUG,
theu're defined really inconsistent acros CMake and SCons.
Used more reliable NDEBUG definition for IDP_spit.
The issue was caused by the wrong attributes maps in certain
circumstances after union intersections.
Namely issue might have happen when more than one iteration of
union was happening and it was caused by the fact that new faces
might be allocated on the same address as freed face from the
old mesh.
Didn't find a nicer fix for this apart from correcting the whole
attributes map after each union step.
We could try removing attributes for the meshes which are getting
deleted, but in asymptotic it's gonna to give exactly the same
complexity as the current approach.
This commit merges the code in the pie-menu branch.
As per decisions taken the last few days, there are no pie menus
included and there will be an official add-on including overrides of
some keys with pie menus. However, people will now be able to use the
new code in python.
Full Documentation is in http://wiki.blender.org/index.php/Dev:Ref/
Thanks:
Campbell Barton, Dalai Felinto and Ton Roosendaal for the code review
and design comments
Jonathan Williamson, Pawel Lyczkowski, Pablo Vazquez among others for
suggestions during the development.
Special Thanks to Sean Olson, for his support, suggestions, testing and
merciless bugging so that I would finish the pie menu code. Without him
we wouldn't be here. Also to the rest of the developers of the original
python add-on, Patrick Moore and Dan Eicher and finally to Matt Ebb, who
did the research and first implementation and whose code I used to get
started.
Was more a feature request actually, but anyway...
Note this is a bit more complex than it's vgroups counterpart, since we have to handle
relations between keys (relative keys, and position for absolute ones).
This assert happens all the time for character physics in debug mode.
In release mode, the assert is skipped but the code is still incorrect
although it does not cause any crash strangely.
New scons discontinued support of python2.6, so we needed
to build just another python in the release and buildbot
environment.
Hope latest scons upgrade a least bring new msvs support
and not only lead to just-another-frustration.
There might be some more upcoming commits, because you
never can be sure there's no typos in the buidbot script
for until you actually fire the builder up.
Use drop-down menu instead of expanded enum for the view select.
There are usually more views than displays and using drop-down
for display device and expanded enum for view is kindof strange.
- Forgot to handle command line arguments
- Because of the fact we need to be able to
use stdout and stderr we need to use regular
console application for the wrapper.
- Because of using regular application for the
wrapper we need to check forparent PID in the
isStartedFromCommandPrompt().
I really hope it's not gonna to become any more
complicated.
Main idea of this change is to make a small executable which
will set OMP_WAIT_POLICY environment variable and then will
launch blender itself.
This is a teporary solution for the time being we'll have a
more clear solution to high CPU usage with OpenMP when building
with MSVC.
Reviewers: campbellbarton, juicyfruit
Reviewed By: juicyfruit
Differential Revision: https://developer.blender.org/D717
Looks like mask points coloring was recently changed, and IMAGE space colors were left uninitialized...
Factorized a bit the code about vertex_handle & co too, was quite duplicated.
Issue is that, when we switch from full preview view to combined view, preview area has
its `sizey` set to use all available height, which leaves nothing for poor main area.
For now, make checks when switching mode to combined one, so that both areas have
resonable `sizey` values. Not perfect, but works.
Added a small menu with a few helper oerators next to each group panel:
* Remove group from all objects
* Select objects in group
More could be added possibly in the future.
Thanks to Campbell for the advice here.
The issue was that some properties are no direct children of the struct we support in 'copy to selected'
(RNA_Sequence in this case). Since we can't use the ID of sequences here (it's the scene, while we need
a sequence level of control), we had to add a new API helper to RNA path, which takes a RNA type
and return a path relative to the closest ancester of that type.
This way, we get a path from the RNA_Sequence, and can easily apply it to all other valid sequences
to copy the property.
Review, suggestions and edits by Campbell Barton, thanks!
In collaboration with Sergey Sharybin.
Also thanks to Wolfgang Faehnle (mib2berlin) for help testing the
solutions.
Reviewers: sergey
Differential Revision: https://developer.blender.org/D690
The typos didn't cause any bug, but the mis-ordered parameter names in CarveExporter_InitGeomArrays were confusing.
Reviewers: sergey
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D709
For now it was mainly about OpenCL wrangler being duplicated
between Cycles and Compositor, but with OpenSubdiv work those
wranglers were gonna to be duplicated just once again.
This commit makes it so Cycles and Compositor uses wranglers
from this repositories:
- https://github.com/CudaWrangler/cuew
- https://github.com/OpenCLWrangler/clew
This repositories are based on the wranglers we used before
and they'll be likely continued maintaining by us plus some
more players in the market.
Pretty much straightforward change with some tricks in the
CMake/SCons to make this libs being passed to the linker
after all other libraries in order to make OpenSubdiv linked
against those wranglers in the future.
For those who're worrying about Cycles being less standalone,
it's not truth, it's rather more flexible now and in the future
different wranglers might be used in Cycles. For now it'll
just mean those libs would need to be put into Cycles repository
together with some other libs from Blender such as mikkspace.
This is mainly platform maintenance commit, should not be any
changes to the user space.
Reviewers: juicyfruit, dingto, campbellbarton
Reviewed By: juicyfruit, dingto, campbellbarton
Differential Revision: https://developer.blender.org/D707
When menus are clamped to the window bounds,
its was possible not to have an active menu-item under the mouse,
Making Ctrl+S,Enter not completely reliable.
Changes needed to support this are:
- menu item is activated on popup menus
(to avoid relying on mouse-over)
- moving mouse away from menu items only de-activates when over a new menu-item.
- Mouse clicks are ignored if they are not directly over the menu item.
The issue was caused by NLA evaluation without actions not setting
id's flag as updated (as that's happening when action writes data
to the ID datablock).
Added the same flag set for the NLA evaluation as what's happening
for actions.
This is another example of transform code crappiness. Projection
snapping for objects is handled separately than regular snapping.
Luckilly for us, we have the normal ready from the raycast result and a
copy of code from ElementRotation function can do the necessary
rotations for us.
I have not enforced constraints here (arguably, maybe I should, and the
already present projection snapping should do that too but seems it is
commented out and the unorthodox order of operations here has me a bit
scared. Leaving as TODO)
The issue was caused by the render engine loading edit mesh, which re-allocates
mesh array which might be referenced by other object's derived meshed.
Worst thing about this is that updating render engine happens from the end of
scene update function, after all the objects are updated and so. This is needed
so render engine gets the update objects which is correct.
The only proper way to solve the issue is to make it so viewport engine does not
leave objects in inconsistent state, meaning nobody will reference to freed data.
In order to reach this we do edit mesh loading before running objects update so
all the objects which uses that mesh will have proper references in the derived
mesh.
This also solves old creepyness which happened before when having single object
in edit mode. tweaking it will calculate derived mesh as a part of scene update,
then this derived mesh will be freed by edit mesh loading and viewport will be
creating derived mesh again.
Now render engine is expected to do nothing with meshes which are in edit mode,
but they still need to load edit data for non0meshes. It's not really easy to
do from the BKE level because needed functions are implemented in the editor.
Thanks Campbell for the review!
Differential Revision: https://developer.blender.org/D697
Yet another attempt to fix the annoying lines that crop up in texture
painting on edges and quads. I feel this needs better investigation but
using pixel tolerance here looks like alleviates the issue.
(original patch by Sergey Sharybin)
Note: RNA API can't use size_t at the moment. Once it does this patch
can be tweaked a bit to fully benefit from size_t larger dimensions.
(right now num_pixels is passed as int)
Reviewed By: sergey, campbellbarton
Differential Revision: https://developer.blender.org/D688
(original patch by Sergey Sharybin)
Note: RNA API can't use size_t at the moment. Once it does this patch
can be tweaked a bit to fully benefit from size_t larger dimensions.
(right now num_pixels is passed as int)
Reviewed By: sergey, campbellbarton
Differential Revision: https://developer.blender.org/D688
Much better to use small loops when doing complex operations over color elements
(any serious compiler will flatten them anyway), avoids (some!) stupid mistakes when
editing their code.
Also, use min/max funcs instead of lengthier 'if (foo < 0) foo = 0'.
Turns out to be mostly some cleanup in Pose select code, got rid of magic numbers
(now understand usual SEL_xxx enums) in ED_pose_deselectall(), which was renamed
to ED_pose_de_selectall, and have a new bool parameter to ignore visibility status
in its process (was the root of the reported issue).
Also factorized slightly "(de)select all" code. Yet this area could use much more
cleanup probably...
1. This patch fix the KX_ConstraintWrapper documentation (radian instead of degrees).
2. It also adds the missing GENERIC_6DOF_CONSTRAINT constant.
Reviewers: dfelinto
Reviewed By: dfelinto
Differential Revision: https://developer.blender.org/D672
- due OSL i386 never worked on OSX, the new libs do not even contain this arch !
- As we had to fix duplicated symbols from generic UTF finctions same in LLVM and COLLADA,
LLVM-less build must have UTF lib reenabled
Note than it's using nearest faces, since it showed to be much more performant than
nearest vertex (quite odd, it's about 40% slower for the first element, then 50 times quicker
for all others, as if BVH was cached, and building face was slower than verts one,
but then using it, much quicker!).
Couple of issues:
- Debian systems doesn't use site-packages but uses
dist-packages instead.
- Requetss from ubuntu repository depends on urlllib3
and chardet, which also implies six. copy those libs
as well.
I know it's all rather annoying, but no that much
choise is in here..
This solves missing requests package reported on the systems where it's
located in dist-packages rather than in site-packages.
To do this there's now a helper macros which handles both requests and
numpy now and could be used for more packages in the future.
Reviewers: campbellbarton
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D686
Issue was caused by rB47ec0394ca3d, which disabled BBox check in editmode - but bbox check was also
setting `len_diff`, which is mandatory when doing ray_start_local correction for ortho view...
Now, in this case, we do a quick rough compute of len_diff from vertices coordinates (accuracy is not
needed here, we just have to be sure corrected `ray_start_local` remains 'before' (outside) of the
geometry).
Disable adding snapping point outside of 3D space for now,
visualization of the points is not implemented outside of
this space and silently adding them wouldn't really be a
good idea.
That's rather tricky to think of a good threshold here, and
maybe we'd better use something based on the number of faces
instead.
Anyway, let's give it a try this way and see what happens.
Issue wascaused by the famous OpenMP crap in MSVC2013,
so only way is to use openmp threading if number of BVH
nodes is high enough.
Made it 8 for now, which seems to work rather fine on my
laptop and adult dragon from sintel. But maybe it's to be
adjusted a bit more.
In pipeline.c, do_render_3d() is called multiple times for each frame when
motion blur is used. This caused duplicates of the same struct Render instance
in re->freestyle_renders, resulting in fatal double freeing of allocated memory.
uses old behavior again. OSX menu and CTL-CMD-F still work as lion fullscreen as well as right-upper corner fs window-icon
- We must investigate here why double promotion happens from op calls ( dispatchEvents on redraw cause duplicated calls here )
- The actual op calls cause fs to be in a wrong state, so also mousehandles fail and CTX_wm_window(C) is not valid.
- similar problem is with quit op, which does not close the app right ( totblocks )
- i would prefer to try getting direct os function call here rather
This allows adding a "fake" sun beam effect, simulating crepuscular rays
from light being scattered in a medium like the atmosphere or deep water.
Such effects can be created also by renderers using volumetric lighting,
but the compositor feature is a lot cheaper and is independent from 3D
rendering. This makes it ideally suited for motion graphics.
The implementation uses am optimized accumulation method for gathering
color values along a line segment. The inner buffer loop uses fixed
offset increments to avoid unnecessary multiplications and avoids
variables by using compile-time specialization (see inline comments
for further details).
Baking progress preview is not possible, in parts due to the way the API
was designed. But at least you get to see the progress bar while baking.
Reviewers: sergey
Differential Revision: https://developer.blender.org/D656
This is a much simpler fix to that libyaml-cpp mess, since doing it that way we get our static own
libyaml-cpp... And this should work for all platforms (realized libymal-cpp-ver option was only
implemented for deb-like distro, since it was the first place where issue showed up :/ ).
Also extended the size of buf[] in print_error() to prevent mem_printmemlist_pydict_script[]
from getting truncated when MEM_printmemlist_pydict() is used.
Differential revision: https://developer.blender.org/D675
Reviewed by: Campbell Barton
bmesh's 'Join Edge Kill Vert' was using own code to delete loops, instead of
using `bm_kill_only_loop()`, which did not tag loop indices as dirty (causing
the crash), but also did not free loop's CDData...
In addition to D319, this patch updates the parameter editor, the UI of Freestyle.
Using new API functionality and experience gained in making D319, this patch
provides a quite noticable speedup for commonly-used Freestyle linestyle modifiers.
As this patch touches a lot of code (and mainly the foundations) it is likely that
mistakes are made. The patch has been tested with a regression suite for Freestyle
(https://github.com/folkertdev/freestyle-regression-tests/tree/master), but testing
with scenes used in production is very much appreciated.
Differential revision: https://developer.blender.org/D623
Author: flokkievids (Folkert de Vries)
Reviewed by: kjym3 (Tamito Kajiyama)
Issue with previous code was that branch selected was applied to all repositories.
To be able to control branch on a per-repo basis, we need to use codebases instead.
This commit only enables branch (master/testbuild) and revision selection for master,
we probably do not need this for submodules currently (easy to enable anyway).
Note: tested on local buildbot setup, for linux only (up to build step, have no build env).
If the active image node contributes to the final material shader
(meaning it's either directly or indirectly connected to an Output Node)
the user will receive an alert about circular dependency.
Similar to what we do for Blender internal the baking will still happen,
but the user will receive the alert which should prevent the image
saving to happen if the result was not intentional.
Core function to check for node output written by Lukas Toenne.
Reviewers: lukastoenne, campbellbarton
Differential Revision: https://developer.blender.org/D673
This patch creates an interface for selection mechanisms in opengl. This
makes it possible to switch between occlusion query based or select
rendermode based selection transparently.
This is really useful on graphics drivers that do not accelerate the
select rendermode path (some ATI cards are notorious for this, and the
new path is used by default there), since occlusion queries are always
hardware accelerated due to their use in games.
The option can be found under system - selection. Auto just enables
occlusion queries for ATI users while the rest of the options enforce
one of the two methods always.
There is just one known change, previous code enforced nearest bone to
always get selected, even when mouse selecting near the same position, I
couldn't replicate the behaviour though.
patch by me with edits and review by Campbell.
Thanks!
We do have odd behaviors with name and expanded enums, but in this case it did
not made any sense at all! :)
Note the error was not in RNA, but in C UILayout api itself...
Fix T41115: Motion Blur renders Objects Black - But not in Viewport Preview
This actually extends previous fix to normals and makes it all much nicer now.
Worth doing some intense testing, quick one worked just fine but there always
could be some corner cases.
Rename UI_init_userdef_factory to BLO_update_defaults_userpref_blend
This closely matches BLO_update_defaults_startup_blend so makes sense for them to be together.
Fix T41079: Solid black render of object with negative scale and smooth shading
In both cases the issue was caused by negative scaled objects with single mesh
users for which scale gets applied when using static BVH.
Since the on-fly normals calculation land normals for such cases weren't flipped
leading them to point to a wrong direction.
Added a special object flag for this, which is a bit of a bummer because now
we've got less bits for real useful things, but this is the only way to get
proper normals without adding more complexity in the on-fly calculations.
Previously the imported mesh data were transformed by a local-to-world projection matrix
through NodeTransform. Now that the file loader assumes the imported mesh data already
in the camera coordinate system, an indentity matrix was supplied to NodeTransform.
The present commit avoids this redundant matrix operation by using NodeGroup instead of
NodeTransform, recalling the latter is a subclass of the former with extra matrix computations.
The issue is actually that creating a new image in texture paint mode
will set it always as a stencil image. Internally, the code checks if
the painted image is the same as the stencil and if it is, no painting
is done.
Solution is to expose a boolena to the operator for setting the image as
a stencil (could be an enum in th future for more uses)
Stencil UI is a bit weird here, will definitely redesign.
The button was intended to guide new users of Freestyle, but actually that is a matter
of documentation. The button is unlikely to be frequently used once users get familiar
with Freestyle panels in other contexts. Now that Freestyle has been part of Blender
since a while, it seems time to clean the UI.
We were copying everything from the old sequence into each new ones... including the stripdata,
which for image sequences is an array with one item per image!
So bug was an exponential one, separating strips of a few tens of images was insensible, while
separating a strip of 1000 images would add above 250MB to file size (and RAM usage too)!
The issue was caused by the wrong objects order for
convertblender.c. Dependency graph totally missed
handling of this situation.
Fixed now, but it's not that pretty, don't try this
at home obviously. But consider this to be good enough
for the current dependency graph.
Details are in the build_dag_object().
Commented out the 'no zero' protection of scaling transforms for numinput.
Issue is, once an axis has null scale, you can't regrow it from transform code
(you have to directly edit the scale property). This is not ideal, but getting
good behavior in this case is hairy...
Yet, when using numinput, you type precise values, so if you want to set it to zero,
set it to zero. User is assumed responsible, we should avoid too much 'invisible magic'
when handling precise inputs. ;)
Note: an idea for possible future feature would be to have an 'absolute' mode for numinput
(allowing to type in real value, not factors).
Calling `glLoadName()` was possible with '-1' value (65535 actually), in `draw_pose_bones()`...
Note I think bug exists since ages (at least, seems to be already present in 2008
when drawarmature.c was added in repo)!
Odd behavior reported (issue only affected objects once parented to the armature) was
due to ordering stuff, since only objects drawn *after* the armature were affected.
Yep, at last it's here!
There are a few minor issues remaining but development can go on in
master after discussion at blender institute.
For full list of features see:
http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.72/Painting
Thanks to Sergey and Campbell for the extensive review and to the
countless artists that have given their input and reported issues during
development.
That's really annoying that multiplication order is flipped
comparing mat3 and mat4 cases, but for the purposes of not
breaking all the branches which might use this stuff we'd
better keep order consistent with old version for now.
Suggestion here would be to make order consistent but rename
this functions to mult_* to make compilation fail instead
of failing and using wrong order silently.
The "Show All" and "Show Selected" operators in the Graph Editor was taking into
account all handles on keyframes, even when some of those would be invalid and/or
set to nonsense values (e.g. for any interpolation mode other than "Bezier")
There were several small issues/inconsistencies if how particles' org face index was checked,
leading in some cases to invalid indices and hence mem access, in RNA UV/VCol compute for particles.
Note org code RNA one was copied from (in BI's convertblender.s) is much more complicated,
and seems to never reach those breaking conditions.
Also deduplicated most code in those UV/VCol particles funcs, they were doing mostly the same thing!
Finally, also got rid of annoying `NO CD_ORIGSPACE, error out of range` error message in console,
was another case of not checking whether we did have any faces in final mesh!
Currently this gaussian blur implementation accumulates values in the
square kernel rather that doing X direction and then Y direction because
of the lack of using multiple-staged filters.
Once we can we'll implement a way to apply filter as multiple stages we
can optimize hell of a lot in here.
Another thing we can do is to use SSE2 instructions here.
Manifold was defined as any edges not using 2 faces,
however its useful to have some options here.
You can now select between Wire/Boundary/Multi-Face/Non-Contiguous/Verts
Note the Non-Contiguous option is new,
it selects edges between faces pointing in different directions.
- name primary comparison var 'v'
- names remain same when expanded
- no odd mixing of macros, use prev ELEM + extra arg
- use 16, even if not used yet, saves adding more in future
BKE_scene_base_iter_next() was completely messing poor dupli objects' matrices...
Note this func should be reworked, but as stated in comments, it should not exist at all,
DAG should be used here, so until we have new shinny one we can live with this.
Also, mballs do not behave correctly when used as duplis (org object remains visible/rendered,
unlike any other object type). This will be fixed in a separate patch/commit,
since it proved to be rather tricky to handle.
BKE_mball_basis_find() was making a bunch of string manipulations for all MBall objects
(including duplis), and then making a (broken!) check to do nothing in case of duplis mball!
Now it makes correct check in early stage.
Preset operators should avoid using `bpy.context.object.data` as a base
path to properties. This path is not available in the buttons context
when using pinned datablocks! Instead use the specific
`bpy.context.camera` and `bpy.context.lamp.sky` paths now, which lead to
the correct datablocks in any case.
1. Add attribute to get the constraint type.
2. Add missing documentation for getParent, setParam, constraint_id in bge.types.KX_ConstraintWrapper.rst.
3. Add missing documentation for GENERIC_6DOF_CONSTRAINT and flag bit in bge.constraints.rst.
4. Fix typo in CcdPhysicsEnvironment.cpp
Reviewers: moguri
Reviewed By: moguri
Differential Revision: https://developer.blender.org/D654
This patch adds to the existing property actuator a level mode, which is switching the property depending on the input level.
Reviewers: moguri
Reviewed By: moguri
Differential Revision: https://developer.blender.org/D652
Particles could completely cancel Hide flag! 'Accumulative' bool is not a really good idea here,
hide (or render-hide) are some kind of 'absolute' no-go.
Found another issue in that area, duplicated objects would still show in 'render override' mode,
when object was render-disabled.
Hopefully things are better now.
Note: This makes no change to the user, the render engine (Cycles) still
need to update the progress during baking. But this is the Blender side
of this.
In collaboration with Benoit Bolsee (mainly doing it under his
directions).
Note: FFmpeg lib needs to be compiled with rtsp support for this to
work.
Bug 1/2 of T41004
Now the bevel tool, modifier, and internal operator have a material
slot # parameter that the user can set. If left at default of -1,
behavior is as current -- bevel face material is taken from the
closest original face (this may be ambiguous). If material slot
is >= 0, it gives the material slot index number for the material
to use.
- add BLI_stack_count
- add BLI_stack_pop_n to pop into an array
- add BLI_stack_push_r, which returns a pointer that can be filled in
Also remove sanity check in BLI_stack_pop, assert if the stack is empty.
trees for localization (previews and viewer evaluation).
This is handled entirely by the compositor already. Doing this during
localization is redundant and risks divergent behavior.
Proxy operations from muted nodes would still create conversion
operations where the datatypes don't match, which creates unexpected
behavior. Arguably datatype conversion could still happen even when the
main operation is muted, but this would be a design change and so is
disabled now.
This is related to Task T34861 to increase up & track axis options for TrackTo actuator. I've just added it to differential to facilitate an easier review.
With the patch applied you can select X, Y and Z axis for the Up axis, and X, Y, Z, -X, -Y and -Z for the track axis.
Related to the implementation I have used the algorithm from Trackto constrain placed in constrain.c but adapted to be used with MOTO library.
The wiki docs are here (http://wiki.blender.org/index.php/User:Lordloki/Doc:2.6/Manual/Game_Engine/Logic/Actuators/Edit_Object#Trackto_Actuator).
Test file is here: {F97623}
I have also uploaded 2 screenshots showing the UI modifications to the TrackTo actuator:
{F91992} {F91990}
Reviewers: moguri, dfelinto
Reviewed By: moguri
CC: Genome36
Differential Revision: https://developer.blender.org/D565
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.
This was a ToDo item, for mesh-based rigid body shapes (trimesh, convex)
the operator was simply using the bounding box volume, which can grossly
overestimate the volume and mass.
Calculating the actual volume of a mesh is not so difficult after all,
see e.g.
http://research.microsoft.com/en-us/um/people/chazhang/publications/icip01_ChaZhang.pdf
This patch also allows calculating the center-of-mass in the same way.
This is currently unused, because the rigid body system assumes the CoM
to be the same as the geometric object center. This is fine most of the
time, adding such user settings for "center-of-mass offset" would also
add quite a bit of complexity in user space, but it could be necessary
at some point. A number of other physical properties could be calculated
using the same principle, e.g. the moment of inertia.
This is mostly the same fix as before, but now code depending on culling
checks is executed after KX_Scene->CalculateVisibleMeshes(). As a
side-effect, LoD checks and animation culling now use the current
frame's culling information rather than the previous frame's.
Fix T41013: OSL and Crash
Fix T40989: Intermittent crash clicking material color selector
Issue was caused by not enough precision for inversion threshold.
Use double precision for this threshold now. We might want to
investigate this code a bit more further, stock implementation
uses doubles for all computation. Using floats might be a reason
of bad rows distribution in theory.
This way no access to automated tests variables will happen.
Ideally it should be something more like using blenderlite
with just Cycles enabled, but let's keep it for a bigger
clean up.
We should do autotests only after setting path to sconsign,
otherwise using the same working directory and different
build directories will eventually break.
Clipping border causes a software fallback on ATIs. We have disabled it
in that case but from minor digging on the net seems like Intels support
this, so enable.
Issue here is that we force mirroring even if original armature is not
mirrored.
We could be smart and store only unselected mirrored bones here (since
those will get restored from transdata), however not all properties were
getting stored and restored; rolling bones still suffered from the bug
for instance.
To fix this we need to restore all properties that armature mirroring
overrides. Transdata obviously does not offer a lot of space here, so I
used TransInfo->customdata to store an array of initial parameters of
the mirrored bones.
material is applied to multiple Dupli instances of an object.
One of the random_id initialization lines for cycles objects slipped
into the basic update part in this commit:
rBb98ff5cb5b2c14c33b16e3b129e1e08810e90a6c
This would constantly re-shuffle the random_id ...
There were some missing updates in the viewport render job which lead to
wrong SSS mapping on the final resolution.
There was also wrong scaling applying when border render is used.
And last but not least(?) strands render was using first level of the
resolution leading to really thick strands in the final viewport.
correctly.
Problem was that object layers are defined by duplis as the top-level
duplicator layers. This happens //during// the duplilist construction,
which breaks group layer checks for subsequent instances and hides them.
Now the duplilist generators leave Object DNA untouched, the
modification of layers for drawing, rendering, etc. happens afterward
in the duplilist_apply/restore functions, as a kind of second pass.
To do so, matching BKE 'API' was also refactored a bit:
* Get Pose data instead of Object, as parameter;
* Removed some sanity checks not needed at such a low level (callers are supposed to do that);
* You can now remove an arbitrary bone group, not only the active one.
Based on patch by pkrime (Paolo Acampora), with own edits.
Reviewers: #python, pkrime, aligorith
Reviewed By: aligorith
Differential Revision: https://developer.blender.org/D522
- this workaround makes sure the breaking -fopenmp is not exposed to testcompiles which use C_FLAGS respective CXX_Flags
- -fopenmp is always added to the CMAKE_C(XX)_FLAGS_<BUILD_TYPE> now
- another way to solve this would be to move the Configure OpenMP part after all line 2366,
but i don’t know if this would be apreciated
Viewers were activated both inside the active group as well as the top
level tree (the latter being a quick fix for getting a fallback viewer).
This caused a race condition on the shared viewer image.
Now the active viewer is defined at node conversion time in the converter
so that only one can be active at a time without each node having to
follow complicated rules for exclusion.
- test fail due libiomp5 is not available @ testing time
- must solve this by adding requred flags/includes to tests
- atm release builds are possible with this hack
This is related to task T29419. Credit also goes to Goran Milovanovic
(goran) for proposing an initial fix for this issue.
The issue is the current behavior of applyImpulse doesn't match the behavior
described in the documentation as instead of a impulse point in world coordinates,
it seems to require a coordinate in a local space.
Additionally, applyImpulse function isn't consistent with similar functions (applyForce, applyTorque, etc)
as it doesn't allow to choose in which space (local or global) the impulse is applied.
Now, we have the following function:
applyImpulse(point, impulse, local=False)
being "point" the point to apply the impulse to (in world or local coordinates). When local is False will
have both point and impulse in World space and when local is True will have point and impulse in local space.
Reviewers: moguri, dfelinto, brita_
Reviewed By: moguri
Differential Revision: https://developer.blender.org/D567
Fix T39286: Display percentage ignored in Cycles viewport.
The threaded depsgraph update changes included a cleanup of the global
is_rendering flag, which was replaced by a general EvalContext being
passed to dupli functions.
Problem is that the global flag was true for viewport duplis before
(ugly hack), which was used as a check for generating dupli orco/UV from
mesh data layers. The new flag is stricter and only true for actual
renders, which disables these attributes and breaks the Cycles
Texture Coordinates and UVMap nodes.
The solution is to extend the simple for_render boolean to an enum:
* VIEWPORT: OpenGL viewport drawing (dupli tex coords omitted)
* PREVIEW: Viewport preview render (simplified modifiers)
* RENDER: Full render with all details and attributes
There are still some areas that need to be examined, in particular
modifiers seem to totally ignore the EvaluationContext!
Instead they generally execute without render params from the depsgraph
(BKE_object_handle_update_ex) and are built with render settings
explicitly.
Differential Revision: https://developer.blender.org/D613
New properties 'line_color' and 'line_priority' are added to Material ID data blocks.
The 'line_color' property allows users to specify a per-material line color that can be
used as a Freestyle line color through Material color modifiers of line style settings.
The new line color property is intended to provide a solution for line color
stylization when a proper Freestyle support for Cycles is implemented (likely
as part of the upcoming Blender 2.72 release; see Patch D632). Materials in
Cycles are usually set up using shader nodes, and Freestyle won't be capable
of retrieving colors and other properties from node-based materials any soon.
The new line color property of materials addresses this foreseen limitation by
providing artists with an intuitive alternative mean to specify line colors on a
per-material basis independently from node trees.
The 'line_priority' property gives users a way to control line colors at material
boundaries. When a line is drawn along a feature edge at material boundaries,
one of the two materials on both sides of the edge has to be picked up to
determine the line color. So far there was no way to control this selection
(which was in effect at random). Now the material with a higher line color
priority will be selected.
The new per-material line settings are shown in the new Freestyle Line tab in
the Material context of the Properties window (only when Freestyle is enabled).
This can be used to subtract a region from masks but it's not very
user friendly yet. To male this work better, the tool will probably be
brushified later.
* malloc() is used now, which is supported since sm_20: http://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#dynamic-global-memory-allocation-and-operations The performance of this needs to be tested on various cards still.
* This also works for Heterogeneous Decoupled Ray Marching, but in this case I get sporadic "Illegal Address" errors on my Geforce 540, therefore I did not remove the GPU check in kernel_volume_use_decoupled() yet.
I would appreciate some tests from people who compile themselves, enable Volumetrics in kernel_types.h.
Handle vectors for the initial location were only stored when the handles are actually
selected for transform. When only the master vertex is moved this can still affect
the handles though, since they are constraint to one side of the master vertex.
Now the handle data is stored regardless of whether handles are included in the transform,
so on cancel their location is reset correctly as well.
It is sometimes possible to end up with a lot of datablocks which have old + unused
"AnimData" containers still attached. This most commonly happens when doing motion
graphics work (i.e. when some linked-in objects may have previously been used to develop
a set of reusable assets), and is particularly distracting in the NLA Editor.
This commit adds an operator which removes AnimData blocks (restricted to only those
which are visible in the animation editor where it is run from) which are "empty"
(i.e. that is, have no active action, drivers, and nla tracks or strips).
This operator can be found from the "Edit" menu in the NLA Editor. Although it also
works when run from the DopeSheet or Graph Editors, it is of less use there since
those won't show these empty AnimData blocks by default (since by definition, such
AnimData blocks necesarily have no keyframes or drivers that can be shown), hence
there will be no feedback if the operator fails or succeeds.
* Put Normal Settings beneath the other ones, wild button jumping should be avoided.
* Remove Cage prefix for Object and Extrusion, it's clear from the button placement, the former UI was too squeezed...
* CUDA can be compiled with Volume support again, change line 78 kernel_types.h for that.
Volumes are still fragile on GPU though, got some Memory/Address CUDA errors in tests.. needs to be investigated more deeply.
Ne op is similar to other 'Select Grouped' ones in Blender.
Based on patch D288 by Cruentus_Nex (Steve).
Reviewed by Campbell Barton, thanks!
Revision: D288.
Noticed this while looking into something else.
The change is trivial, but gives a rather nice preformance improvement,
so why not.
Theres's actually a lot one can do to improve collision performance if
one wanted to, the triangle-triangle check alone has a lot of room for
improvement.
* Allows drag and place workflow in addition to click workflow
* Should be compatible with future use of calling operator and placing
instead of left-clicking
In rB78c491e the `initialize_particle` function was split into 2 parts for particle texture initialization.
The texture init part however also initializes birth times, which is now missing in the main init function
in some cases (notably when setting start/end directly without a subsequent time step).
Blender Internal only uses materials with non-zero user counter for SSS and new
mutable libblock copy was keeping users counter at zero.
Now it sets user counter to 1, which is a bit of arguable decision, but which
also kind of makes sense -- meaning callee owns the copied block.
The cause of the crash was identified in an uninitialized member variable
`Main->lock`. Now that struct Main has a few member variables whose
values are dynamically allocated, per-render Freestyle-specific Main data
structures will be allocated and released using `BKE_main_new()` and
`BKE_main_free()`, respectively.
This revision complements the commit rB6135556f4556.
A bit hackish solution for now, cleaner solution we'll look into as a
part of the new DAG project, when it's clear what kind of data is passed
to the evaluation callbacks.
The problem is that the render window keeps keybord input focus even after it has been
lowered. Windows maintains the Z-order of windows, so the present solution is to raise
the window that has been just below the render window.
Differential revision: https://developer.blender.org/D594
Reviewed by: campbellbarton
Probably a leftover from old code, I don't think this invoke usage of VIEW3D_OT_background_image_add
was actually reachable anywhere from the UI, but managed to get it working from py, without increasing
user count...
Preserve buffer form previous runs so it's possible to make
a compo of full frame, then draw a border and start tweaking
nodes and see updates in that border.
Main idea is to make it able to visually compare difference
between what was changed inside the border and how frame
looked before the tweaks outside of the border.
Also implemented Clear Viewer Border in compositor, shortcut
it Ctrl-Alt-B.
Reviewers: lukastoenne, jbakker
CC: venomgfx, sebastian_k
Differential Revision: https://developer.blender.org/D582
Suppress warning for now, it's harmless and only happens
with new libpng.
In the future we might try enabling it for non-datatoc-ed
files, but it's really not worth spending lots of time on.
Volume scatter might happen before path termination, so
need to check transparent bounces and consider shadow an
opaque when max transparent bounces are reached.
TODO: CPU code seems to have different branching in conditions
which made me thinking it does different things with volume
attenuation, but from the render results it seems the same
exact things are happening there. Worth looking into making
simplifying code a bit here to improve readability.
Newest clang would use 'cfe-' prefix instead of 'clang-' one - don’t ask me why...
Had to make a wrapper around wget, that:
* Accepts an array of URLs as first parameter;
* Errors and exits when wget fails for some reason (was silent before).
Issue reported in T40871 by simonrepp (Simon Repp), thanks.
Objects were hidden in "only render" mode if they were duplicators. This is correct in general, but for particles should be disabled by the "show emitter" option.
Simply add an option to render settings to save an EXR cache,
just when the render is finished. Also changed RE_ReadRenderResult() to read
cache instead of temp sample files (those are fully volatile now anyway).
Path to save cached render results is an UserPreferences setting.
Also added 'Reload render' feature to the Image Editor (so one can now re-open a blend,
and in an Image Editor hit ctrl-R to (try to) reload last render from cache).
Reviewers: campbellbarton, sergey
Differential Revision: https://developer.blender.org/D553
This brings new copyright header which supports GPL2 and 3.
It wasn't really an issue before because we had agreement with
Tobias, but now it's all documented in sources.
The issue was introduced by 0f95149 and it only worked before
because default game material alpha blending was set to alpha.
Now it'll check whether material has transparency enabled and
will use alpha blending by default in this case.
Brings new bounds limiting and also prepares build system
for the changes in the upstream.
Namely shared_ptr header and namespace is now being detected
by a build system rather than by hacks in the code.
This commit includes some changes to auto-detection flags
in SCons, presumably adding more consistency there. This
is main changes which are suppoed to be reviewed here.
Reviewers: campbellbarton
Differential Revision: https://developer.blender.org/D581
Some tools used view vector while others used the area normal.
Area Normal -is- useful and it will have to be incorporated to the
system better, but I'd like to do it in a way that is well
collaborating with sculpt plane calculation too, because it will have
to be slightly more expensive, i.e normal calculated before the plane,
never together, so only front to the area normal influence can be kept.
Currently sculpt plane takes into consideration view vector, but that
can produce an unwanted normal/plane at glancing angles on mesh edges.
Another issue that arises here is whether we do an actual calculation of
sculpt normal for each symmetry pass or we just flip the initial (former
is more expensive but more correct)
objects.
Flush edits for all objects, not just the active one. Here we might want
to disallow leaving an object on sculpt mode when selecting another, but
this works, no need to enforce it.
This way it's not needed to include BLI_threads.h from the
BKE_main.h which helps avoiding adding PThreads includes to
each library which uses Main on Windows.
From the API point of view it's now MainLock* and to lock or
unlock the main you're to use BKE_main_(un)lock().
This solves compilation error on Windows with SCons.
This is a misnomer: the operator copies data (mostly pointers) of
various types from the active to selected objects. "Make Link" implies
creating a permanent connection between these objects ...
Most of the default builtin Keying Sets (e.g. Loc, Rot, Available, etc.) depend
on the currently selected objects, making them unsuitable for use with this feature.
As a result, this feature would silently fail. This commit adds some error messages
which at least let users know what's going on so that they can make other plans
Disclaimer: The author of this patch is Geoffrey Gollmer (gomer). I only updated the patch to the current git master status, reworked several parts to fit well with current coding style and applied several fixes.
This actuator allows users to show/hide the mouse cursor using logic bricks, as well as control object rotation with a mouse in the BGE.
The mouse rotation is flexible enough to allow any type of mouse look, as well as banking for flight controls.
{F94520}
{F91859}
Blend file for testing Mouse actuator (with default parameters and crosshair): {F94920}
Reviewers: moguri
Reviewed By: moguri
CC: gomer, lordodin
Differential Revision: https://developer.blender.org/D559
it was caused by ee5f432. Not sure why exactly blender runs into
threading conflicts here since material preview works.
Let's keep it simple for now and disable material nodes preview,
because it's not supported by the interface anyway.
This is to be included to the final release.
Added a lock to the Main which is getting acquired and released
when modifying it's lists.
Should not be any functional changes now, it just means Main is
now considered safe without worrying about locks in the callee.
- left aligned
- higher contrast between tip text and py-text
- use monospace for py-text
D611 by Severin, design by Plyczkowski, with own minor changes.
This is needed because render threads would do some allocations like
part buffer allocation and so. This is likely harmless with the lock
free allocator (not on Windows tho), but when using guarded allocator
we need to be sure access to the list of MemHead is safe.
This revision is meant to update Freestyle's Python scripts to make full usage
of the new features of Python and Freestyle's Python API.
Freestyle's Python scripts are pretty old already, and were never given much
attention. With the 2.7x generation of Blender coming up, this is an excellent
time to update Freestyle's Python scripts, hopefully adding some new features
and achieving some speed improvements on the way.
Main goals:
* use for loops where possible
* general cleanup, making use of more recent python features (generators,
ternary operator, ect.)
* update the documentation on the way (it's lacking atm)
Differential revision: https://developer.blender.org/D319
Author: flokkievids (Folkert de Vries)
Reviewed by: kjym3 (Tamito Kajiyama)
Don't know when this sneaked in, I did wrote that part of the patch on win VM with MSVC2013... :/
Note: letting asside warnings for now, then should not prevent building anyway.
panels.
Patch by @Severin (with minor modifications by me). As discussed in
{D535} the node editor does not have real modal operator tools like the
3D view or image editors for instance, so it makes sense to utilise it
this way. Tabs really help in this area due to the large amount of node
types and categories. Further tweaks could be made later if the need
arises.
Current temporary data of Blender suffers one major issue - default 'temp' dir on Windows is never
automatically cleaned up, and can end being quite big when used by Blender, especially when we have
to store per-process data (using getpid() in file names).
To address this, this patch:
* Divides tempdir paths in two, one for 'base' temp dir (the same as previous unique tempdir path),
the other is a mkdtemp-generated sub-dir, specific to each Blender instance.
* Only uses base tempdir when we need some shallow persistance accross Blender sessions - and we always
reuse the same filename (quit.blend...) or generate small file (crash reports...).
* Uses temp sub-dir for heavy files like pointcache or renderEXRs (Save Buffer option).
* Erases temp sub-dir on quit or crash.
To get this working it also adds a working 'recursive delete' to BLI_delete() under Windows.
Note that, as in current code, the 'recover render result' hack-feature that was possible
with SaveBuffer option is still removed. A real renderresult cache feature will be added
soon, though.
Reviewers: campbellbarton, brecht, sergey
Reviewed By: campbellbarton, sergey
CC: sergey
Differential Revision: https://developer.blender.org/D531
This way we only catch exceptions which we're intended to
and if something bad happens (like missing import due to
refactor or so) we'll see them instantly.
* Fixed different not-in-sync #ifdef blocks for struct stat variants under Windows.
Comments have been left to indicate the portions of BLI_fileops.h and
BLI_fileops_types.h that need to stay in sync.
* Added BLI_wstat() to de-duplicate #ifdef blocks for stat() variants on Windows.
* Fix for opendir() and associate functions in MinGW not working properly with
non-ASCII, MBCS-compatible paths.
MinGW (FREE_WINDOWS) has opendir() and _wopendir(), and only the
latter accepts a path name of wchar_t type. Rather than messing up with
extra #ifdef's here and there, Blender's own implementations of opendir()
and related functions are used to properly support paths with non-ASCII,
MBCS-compatible characters.
Tested with MSVC 2013 Express, MinGW32 (gcc 4.6.2) and MinGW-w64 (gcc 4.7.1).
Differential Revision: https://developer.blender.org/D605
Reviewed By: campbellbarton
It turns out that the new Beckmann sampling function doesn't work well with
Quasi Monte Carlo sampling, mainly near normal incidence where it can be worse
than the previous sampler. In the new sampler the random number pattern gets
split in two, warped and overlapped, which hurts the stratification, see the
visualization in the differential revision.
Now we use a precomputed table, which is much better behaved. GGX does not seem
to benefit from using a precomputed table.
Disadvantage is that this table adds 1MB of memory usage and 0.03s startup time
to every render (on my quad core CPU).
Differential Revision: https://developer.blender.org/D614
Returning to object mode reactivates any generative modifiers
and this can lead to a polycount explosion.
For now just improve the warning when entering dyntopo with generative
modifiers. I would like to add the ability spawn options to apply or
remove the modifiers too, however separate undo stack system comes back
with a vengeance here, since it won't allow restoring the application/
also may invalidate all sculpt undo in the undo stack prior to the
application (needs investigation).
Currently resolution divider is not exposed to the
interface yet, and i'm not even sure it needs to be
exposed because it's somewhat weird configuration.
Need to check how often artists are changing start
resolution in Cycles.
Pretty much straightforward implementation with the
only weak part: render result is getting re-allocated
and upscaled when current resolution is finished.
Not sure how to make it faster actually. Maybe it's
just a matter of making upscale fast enough.
Needed to fix some possible memory leak happening
in Freestyle when canceling rendering on a special
stage -- it was missing temp bmain free,
Reviewers: campbellbarton, dingto
CC: sebastian_k, fsiddi, venomgfx
Differential Revision: https://developer.blender.org/D609
Needed to make the blender link libraries a global property
now that tests are parallel to source directory.
Current sort order for blender link libraries doesn't work
for tests that start with few defined symbols. Doubling the
lib list works, but a TODO to find a better way (probably
using CMake's own mechanism for tracking dependencies).
** TO BE PORTED BACK TO 2.71 **
As pointed out by Thomas Beck (plasmasolutions), the current behaviour and/or
default values for their parameters didn't quite make sense:
1) Back Easing - The old default value of 0.0 results in some overshoot being applied,
while trying to tweak it up or down resulted in some odd jumps and discontinities.
I've ended up removing some code here which forcibly using a "back" value of 1.7
when users wanted 0.0 instead. There doesn't seem to be any good reason for this.
To ensure that there is still an effect initially, keyframes now get created
with back set to 1.7
2) Elastic Easing - The old default settings of <amplitude = 0, period = 0> resulted
in a curve without any elastic bounce, which wasn't very useful for motion graphics.
Now, default values of amplitude = 0.8 and period = 4.1 get set. These were hand picked
by Thomas to work well when the duration of the motion is 10 frames long (i.e. the
typical length of such effects when doing motion graphics).
Issue here is that if there's a texture in the tree, chances are it has
already been set as active texture so groups are never traversed.
Now changed logic so that if a group node is active, its own active
texture takes priority over the parent group active texture.
Cleanest way here is not do bounding box collision for editmeshes at
all. Decision is taken because:
* Usually we want to do the snapping to the edited mesh anyway (when we
don't the mesh is skipped completely, so we don't need to worry for
extra checks)
* Bounding box is calculated from derived mesh. This means that for
subsurfed meshes for instance, the bounding box may be significantly
smaller than the size of the edit mesh.
Looks like the cleanest way to handle this is to no do bounding box collision
for edit mode at all. But this is easy to enforce
This reverts commit 7b5fe4f316.
Conflicts:
source/blender/editors/transform/transform_snap.c
This is a temporary solution in order to get at least
rest of the blender begin up-to-date on the buildbot.
To be able to compile cubins again we need to switch
OSX builder machine to OSX 10.8 and CUDA toolkit 6,
which might take some time, unfortunately.
Note: the custom UV option is only available when calling the operator
via a script. It's currently not exposed in the UI since it's intended
to be used by scripters
To test it:
bpy.ops.object.bake(type='UV', use_clear=True, uv_layer='MyNewUV')
Reviewers: campbellbarton
Differential Revision: https://developer.blender.org/D546
When converting text to value, units' "value" had only 6 digits of precision,
leading to annoying loss of precision esp. when mixing big and small units
(like e.g. miles and inches).
Exposes all supported unit systems & types, and to_value()/to_string() functions.
Reviewed and enhanced by CampbellBarton, many thanks!
Differential Revision: https://developer.blender.org/D416
This way util_simd.cpp would not require modifications
if/when SSE2 is suddenly supported on 32bit platforms.
This also allowed to unleash some issues with util_simd.h
related on the fact that there size_t and int are actually
the same types.
When autosnap mode was nearest frame or nearest marker, this was incorrectly
converting the units to seconds, making this display unusable for anything.
This revision extends the Freestyle Python API to make for style module writing
easier.
- freestyle.types.Stroke: A proper support for reversed() is implemented. It
works the same with other Python sequence objects (returns an iterator starting
from the end). This is in effect equivalent to Stroke.stroke_vertices_end().
- freestyle.types.StrokeVertexIterator: An incremented, decremented and reversed
method are added. The first two methods return a new StrokeVertexIterator
object that has been incremented and decremented, respectively. The reversed
method returns a new StrokeVertexIterator object that will traverse stroke
vertices in the opposite direction.
- freestyle.types.Interface0DIterator: Its constructor now accepts a Stroke
object to create an Interface0DIterator that traverses stroke vertices. This is
in effect equivalent to Stroke.vertices_begin(). The new API makes stroke
shaders involving function calls much simpler as illustrated below:
# in the old API
it = stroke.stroke_vertices_begin()
for vert in it:
result = somefunc(Interface0DIterator(it))
# in the new API
it = Interface0DIterator(stroke)
for vert in it:
result = somefunc(it)
Differential Revision: https://developer.blender.org/D545
Reviewers: kjym3
If the side of a beveled edge hit another vertex, the offset
amount reset to zero. This was the result of commit rB1582dd5e4d7c
which clamped the amount to zero to avoid creating spikes with
obtuse angles. Now we clamp the amount to the closest end of
the edge to where the amount wants to be.
Also fixes the first part of T40365.
This is needed for popups to chance state once activated,
currently it makes use of operators `check` callback, after values are modified,
as the file selector does already.
Most weight tools also work in edit mode.
This change exposes all applicable tools
within a separate weight tool panel
in the tools tab of the tool shelf
Reviewers: campbellbarton
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D592
* Anisotropic BSDF now supports GGX and Beckmann distributions, Ward has been
removed because other distributions are superior.
* GGX is now the default distribution for all glossy and anisotropic nodes,
since it looks good, has low noise and is fast to evaluate.
* Ashikhmin-Shirley is now available in the Glossy BSDF.
* Ashikhmin-Shirley anisotropic BSDF was added as closure
* Anisotropic BSDF node now has two distributions
Reviewers: brecht, dingto
Differential Revision: https://developer.blender.org/D549
Samples render slower than before, but hopefully this is made up for with
reduced noise in most cases. The main slowdown comes from samples that would
previously be wasted and turn out black, which are now continued.
GGX sampling is about the same speed as before, while for Beckmann it is slower
still. Perhaps optimizations are still possible there, but didn't find anything
easy.
Code from this paper, which comes with sample code:
Importance Sampling Microfacet-Based BSDFs using the Distribution of Visible Normals.
E. Heitz and E. d'Eon, EGSR 2014
Differential Revision: https://developer.blender.org/D572
This gives you "Multiple Importance", "Distance" and "Equiangular" choices.
What multiple importance sampling does is make things more robust to certain
types of noise at the cost of a bit more noise in cases where the individual
strategies are always better.
So if you've got a pretty dense volume that's lit from far away then distance
sampling is usually more efficient. If you've got a light inside or near the
volume then equiangular sampling is better. If you have a combination of both,
then the multiple importance sampling will be better.
* Volume multiple importace sampling support to combine equiangular and distance
sampling, for both homogeneous and heterogeneous volumes.
* Branched path "Sample All Direct Lights" and "Sample All Indirect Lights" now
apply to volumes as well as surfaces.
Implementation note:
For simplicity this is all done with decoupled ray marching, the only case we do
not use decoupled is for distance only sampling with one light sample. The
homogeneous case should still compile on the GPU because it only requires fixed
size storage, but the heterogeneous case will be trickier to get working.
Simple search for intersections became slow for larger concave ngons (100+)
Tested to work with ngons up to 75k sides, performance is approx ~6x faster then scanfill.
This is a 2D version of BLI_kdtree with modifications:
- nodes can be removed
- an index -> node map is stored (especially for tessellation)
* Added support for uchar4 attributes to Cycles' attribute system.
* This is used for Vertex Colors now, which saves some memory (4 unsigned characters, instead of 4 floats).
* GPU Texture Limit on sm_20 and sm_21 decreased from 95 to 94, because we need a new texture for the uchar4 attributes. This is no problem for sm_30 or newer.
Part of my GSoC 2014.
This kernel is compiled with AVX2, FMA3, and BMI compiler flags. At the moment only Intel Haswell benefits from this, but future AMD CPUs will have these instructions as well.
Makes rendering on Haswell CPUs a few percent faster, only benchmarked with clang on OS X though.
Part of my GSoC 2014.
Instead of pre-calculation and storage, we now calculate the face normal during render.
This gives a small slowdown (~1%) but decreases memory usage, which is especially important for GPUs,
where you have limited VRAM.
Part of my GSoC 2014.
This makes the code a bit easier to understand, and might come in handy
if we want to reuse more Embree code.
Differential Revision: https://developer.blender.org/D482
Code by Brecht, with fixes by Lockal, Sergey and myself.
This gives around 30% of speedup for gaussian blur node.
Pretty much straightforward implementation inside the node
itself, but needed to implement some additional things:
- Aligned malloc. It's needed to load data onto SSE registers
faster. based on the aligned_malloc() from Libmv with
some additional trickery going on to support arbitrary
alignment (this magic is needed because of MemHead).
In the practice only 16bit alignment is supported because
of the lack of aligned malloc with arbitrary alignment
for OSX. Not a bit deal for now because we need 16 bytes
alignment at this moment only. Could be tweaked further
later.
- Memory buffers in compositor are now aligned to 16 bytes.
Should be harmless for non-SSE cases too. just mentioning.
Reviewers: campbellbarton, lukastoenne, jbakker
Reviewed By: campbellbarton
CC: lockal
Differential Revision: https://developer.blender.org/D564
This means packed images and movies are now supported when using OSL
backend for material shading.
Uses special file name to distinguish whether image is builtin or not.
This part might become a bit smarted or optimized a bit, but it's good
enough with this implementation already.
With very small meshes or very small bevel amounts, the bevel
profile would be flat even if a round one was requested.
Problem was that the code was checking the length of a cross
product for closeness to zero to test coplanarity. Needed
to normalize things before making that test to account for scale.
Suggestion by Andy Davies (metalliandy) to conform with industry standard (custom cage is something else apparently)
Note: this is the last bake related commit I plan for 2.71/rc (unless
everyone agrees that we could squeeze in D546 - custom UVs, which would
be really nice to add for 2.71 scripters)
Note 2: I'll update the wiki docs shortly
slightly outside the mesh.
Reported by Thomas Beck on irc. Issue here is that the mesh bounding box
changes as we are transforming the vertices. Solution is to collide
against the initial bounding box. Unfortunately the snapping functions
are made in a way that a lot of code needed to be tweaked here, but the
change should be straightforward and harmless (famous last words, I
know).
Ideally we might want to even increase the size of the bounding box a
little (as seen in screen space) to allow snapping even in cases where,
cursor is slightly outside the bounding box, but since this is not so
straightforward to do for all cases, at least for me, leaving this as
a TODO.
In the future i'd rather have this reported to an
upstream instead of adding local changes. It's really
easy to override this changes if patchset is not added
and this is to be fixed in upstream. Also the function
was never used so it was rather totally harmless warning
for us.
There is a new option to select whether you want to use cage or not.
When not using cage the results will be more similar with Blender
Internal, where the inwards rays (trying to hit the highpoly objects)
don't always come from smooth normals. So if the active object has sharp
edges and an EdgeSplit modifier you get bad corners.
This is useful, however, to bake to planes without the need of adding
extra loops around the edges.
When cage is "on" the user can decide on setting a cage extrusion or to
pick a Custom Cage object. The cage extrusion option works in a
duplicated copy of the active object with EdgeSplit modifiers removed to
inforce smooth normals. The custom cage option takes an object with the
same number of faces as the active object (and the same face ordering).
The custom cage now controls the direction and the origin of the
rays casted to the highpoly objects. The direction is a ray from the
point in the cage mesh to the equivalent point to the base mesh. That
means the face normals are entirely ignored when using a cage object.
For developers:
When using an object cage the ray is calculated from the cage mesh to
the base mesh. It uses the barycentric coordinate from the base mesh UV,
so we expect both meshes to have the same primitive ids (which won't be
the case if the cage gets edited in a destructive way).
That fixes T40023 (giving the expected result when 'use_cage' is false).
Thanks for Andy Davies (metalliandy) for the consulting with normal
baking workflow and extensive testing. His 'stress-test' file will be
added later to our svn tests folder. (The file itself is not public yet
since he still has to add testing notes to it).
Many thanks for the reviewers.
More on cages:
http://wiki.polycount.com/NormalMap/#Working_with_Cages
Reviewers: campbellbarton, sergey
CC: adriano, metalliandy, brecht, malkavian
Differential Revision: https://developer.blender.org/D547
This reverts commit da7bdf1b47.
This broke compilation in OSX and some Linux boxes.
Error: Policy "CMP0043" is not known to this version of CMake
Shall we leave those for after RC (or even 2.71) is out? ;)
!!! ANIMATORS/RIGGERS PLEASE TEST !!!
I've reduced the size of the threshold for the keyframe lookup here. This threshold
determines the minimum time in frames between keyframes (i.e. "how close" to each
other they can get). Making this too small causes problems like T39207, but it seems
that the threshold we've been using makes it impossible to get accurate behaviour on
driver curves with keyframes, when the driver target only moves 2cm (i.e. 0.02 BU).
So far, all of the test cases from T39207 seem to work fine, as well as Caminandes 2
files, and Kenny the Caterpillar. The Kiribati rigs/shots (thanks jpbouza for helping
to check on these!) also seem to be fine.
When doing a 'weld' type join where there are two non-beveled edges
in the same plane one beveled one but not the other, then there
should be a curved profile; bug was creating a straight one.
No idea why this issue hasn't been spotted before. Took several hours to
figure out where exactly wrong memory access happens..
P.S. I really want to switch ImBuf->rect from int* to unsigned char*...
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.
We need to 'reset' mouse coordinates to the one it was when the gesture handling started,
else org coords are where the tweak event is created, which gives a noticeable gap
(several pixels) and unwanted behavior like the one retported about file box selection.
Skip doing particle update in object_handle_update if object is in
edit mode.
Object will be re-evaluated on exit from edit mode anyway, so it's
_expected_ to be a safe change.
The root of the issue goes to the fact that we only can
read RGB EXR files, but they could be YCbCr or just Luma.
Added support for this two cases.
Note: internally EXR would still be 3 channels, so no
big memory save would happen here, at least yet.
It's not that trivial to support disapearing tracks in the sequence,
and since we're having ongoing 2D stabilization rework patch wouldn't
want to start doing rather bigger changes here now.
Let's just stick to legacy behavior for this release.
This reverts commits 91429d0, 543ce85
2.70 for non Apple systems.
Also refactored the code that restores the previous openmp thread count.
The logic here was weird, mostly due to all the commit madness with
Apple openmp support. The restored thread count though should not depend
on the on/off state of threaded sculpting (since it has to do with
systems other than sculpting only). For OSX threads are restored to the
system thread count but Jens should recheck here.
Now baking does one AA sample at a time, just like final render. There is
also some code for shader antialiasing that solves T40369 but it is disabled
for now because there may be unpredictable side effects.
Cycles expects to find all node sockets with their correct names, but
this can be changed via the API (see bug report discussion).
Solution for now is to let cycles accept this case gracefully instead
of crashing. The shader will simply use the internal default values for
inputs and any connections will be ignored.
Would be nice to report the error somewhere, but cycles doesn't have a
proper logging system for this purpose yet.
The node was using sampler from the callee node and passed
it to the input nodes. Since the fact that compositor output
node uses NEAREST interpolation (why it uses nearest is the
whole separate story) it's not possible to have subpixel
precision in such cases:
<image> -> <translate> -> <output>
For now solving by hard-coding translate node to use BILINEAR
interpolation. It can't become worse in this node anyway and
the sampling pipeline is to be re-visited from scratch.
This bug fix is likely to affect the visual results of existing Freestyle setups using the
Material color/alpha/thickness modifiers with the 'material_attribute' parameter set to
either 'Diffuse' (default) or 'Specular', and also the Ramp option enabled in the case
of the Material color modifier.
it's possible that runtime optimizer would call get_attribute
with NULL renderstate. As per documentation, it's valid to
return false in that cases and in worst case we'll just miss
some possible optimization.
Supporting such cases would require some bigger changes to
Cycles since attributes are only set to up for the kernel
after shader compilation.
Thanks Brecht for review!
Seems to be caused by the way how the most bottom strip in the stack
used to apply effect. Just rendering strip in this cases will not give
proper results.
Made it so effect is applying between empty imbuf and actual strip.
Seems to work by tests, but more intense testing is required.
Changes were made in Stroke::Resample(int) in C++ to prevent a potential infinite loop
caused by an inconsistency between Stroke::_Length and the stroke length computed
based on stroke vertices. Such a stroke length inconsistency is usually caused by missing
calls of Stroke::UpdateLength() (i.e., API implementation bugs), but also may occur due
to scripting errors in user-defined style modules. This commit is meant to help script
writters to identify the latter error cases. Now Stroke.resample(int) may raise a runtime
error to signal an error condition.
They were using INPUT_SPRING in a way which didn't allow
it to easily redo the operator because INPUT_SPRING internally
is stored as a ration between old value and new one and crease
and bevel were converting this to value delta.
Now made it special input type INPUT_SPRING_DELTA which is
storing delta of the spring, meaning now values in the redo
panel kind of makes sense -- they mean how much to add/remove
to the crease/bevel weight.
Expect to be no functional changes from interactive transform
POV, just a bit more convenient to use redo panel.
The replace mesh actuator was reconverting the mesh data which causes
conflicts with the LoD code. Instead, we just look for an already
converted mesh, which should already be in the scene.
This commit pretty much reverts all the changes related on tile-ability
of the fast gaussian blur. It's not tilable by definition and would almost
always give you seams on the tile boundaries.
Atmind already met the issue and tried to solve it by increasing some
magic constant, which is pretty much likely simply made it so compositor
switched to full-frame calculation in that particular .blend file.
Fast gaussian is really not a production thing and need to be avoided.
We're to improve speed of normal gaussian blur instead.
When you hit "setup tracking scene" in MCE, the active layer usually switches
to the layer where the ground has been created, thus he shadow layer.
New objects are created or appended on that layer which means they will be
default be rendered as shadows only. This behaviour requires muscle memory
and is confusing for new users. This patch is changing the behaviour in a
way that the active layer is remembered from before setting up the scene.
Patch by gottfried, thanks!
Reviewers: sebastian_k, sergey
Differential Revision: https://developer.blender.org/D574
This check prevents using empty (no faces) meshes as rigid bodies.
While the idea makes sense, it also prevents using modifier-constructed
meshes, where faces are added only by the modifiers.
Further the check is very easy to circumvent, by removing faces after
making the rigid body, or by assigning a different mesh datablock
afterward.
Suggested by Fabian Emmes (@der_fab).
The helper function `make_freestyle_edge_mark_hash()` was referring to the
original mesh to determine Freestyle edge marks for individual derived mesh edges.
This is no longer necessary now that derived meshes deliver CD_FREESTYLE_EDGE
and CD_FREESTYLE_FACE layers of their own. The reference of the original mesh
was also inappropriate since the edges coming from one of the operands of a boolean
modifier don't have proper CD_ORIGINDEX values but ORIGINDEX_NONE's.
Many thanks to Sergey Sharybin for patch contributions and discussions.
Used a different technique to resolve "impossible" offset cases
that makes more consistency. Also changed the plane in which
the profile lies for the case with only one beveled edge and
more than 3 other edges.
When "Selected to Active" is not on, we bake all the selected objects.
This is the same behaviour we have for Blender Internal.
Dev note: I moved most of the validation tests to outside the bake()
routine so the function can be called in loop.
Reviewers: campbellbarton
Differential Revision: https://developer.blender.org/D560
Using layer visibility in active render layer makes more accurate
preview but can cause problems in some cases:
https://developer.blender.org/rB1973b17fce65a4dfececb45b19abec37898c1ab5#comment-1
GLSL lamps now ignore layer visibility if lock_camera_and_layers is
OFF or game engine is running. The material lamp group still works
unconditionally though.
This should be the final fix for the applyrotation issue. It baffles me
that the fix involves discarding the scale transformations for the
normals but it works so I'm happy with it.
Reviewers: campbellbarton
Differential Revision: https://developer.blender.org/D554
Transparent objects could become subtly visible by the different sampling
patterns for pixels covered and not covered by the object. It still converged
to the right solution but that can take a while. Now we try to use the same
sampling pattern here.
When rebuilding the polygons that touch bevel-involved vertices,
need to copy the edge attributes from corresponding original edges.
Special treatment of corner segments, to maintain continuity of
smooth and seam attributes.
Another fix: if have four meeting edges, two opposite ones beveled
and the other two not, propgate the non-beveled-edges attributes
across the line that joins them (perpendicular to the bevel).
A node group can have multiple input nodes. In the compositor that means
each of the input sockets has to be connected to the linked outputs,
which is represented by a single link on the outside of the group.
This reverts commit 12abe94de8.
After a long discussion in the bug tracker we decided baking should use
the faces normals for glossy (and combined). This is what Blender
Internal is doing, and one of the more predictable way of yielding
predictable results.
That also means the result will not match the render perfectly, but this
is preferrable over the alternatives at hand.
Conflicts:
intern/cycles/kernel/kernel_bake.h
Looks like the normal update flag is used internally in the modifier
itself. So as a workaround just pass normal update to the nodes when
flood filling
Issue was caused by _wstat returning EOVERFLOW error because
of file size didn't fit into stat structure which was using
long datatype.
The idea of this patch is to use _wstat64 and _stat64 structure
which is capable storing 64bit file sizes.
Made it a typedef for stat structure used by BLI_stat function
in order to make code easier to follow and avoid ifdefs all
over the place.
Additionally solved issue with BLI_exists which was wrongly
returning False in cases destination file is larger then 4GB.
Better fix than rBbef5cb3aa2e5a: consider edges between faces with opposed normals as sharp.
In fact, previous code was broken more deeply in this case (inconsistent normals across
a 'smooth fan') - some loop normals would even never be computed!
Fixing this is possible (even wrote it, actually), but this adds more complexity
to a piece of code that is already awfully complicated, *and* normals in that kind
of smooth fan do not make much sense anyway. So simpler and nicer results with
assuming sharp edges between such 'opposed' faces!
Note that there is some face (loop) ordering black magic at work here, added more comments
to try to explain how and why all this works.
As a bonus, we do not need to check for already computed loop normals anymore, since we
know each 'smooth fan' will be walked once, and only once.
This crash can only happen in case faces in same 'smooth fan' have reversed normals.
To support this, we have to always keep a way to get real values in loop_to_poly,
even when loop itself is tagged as done, it might be needed in computation of one of
its neighbor's split normal later.
Due to skinning changes for multi-threaded animations, the LOD check was
being done after skinning. Now the LOD check is run before animations.
This means the culling information is a frame old, but this should be
fine.
Comments from Brecht Van Lommel:
"""
Currently the viewing direction for each pixel is set to the normal, so
at every pixel glossy is evaluated as if you're looking straight at it.
Blender Internal works the same.
"""
This patch makes baking glossy as viewed from the camera.
Reviewers: brecht
CC: zanqdo
Differential Revision: https://developer.blender.org/D555
The kernel for baking the world texture was the same as the one used for
baking. Now that's separate which allows the kernel to reserve much less
memory.
CMake had this --fast-math flag but scons not, makes a big difference on some
files. Slightly slower rendering might still happen though, but it should not
be this much.
This commit is intended to fully fix the problem described in
https://developer.blender.org/T36425#19 (see also the previous commit rB08528f577dcb).
Addition of a small offset (to avoid singularity in stroke rendering due to overlapping vertices)
was not performed for all overlapping vertices.
Removed the StrokeCleaner and related helper functions which were added as a
temporary workaround in rB2a5b6d9c8f16.
A description of the problem is found in https://developer.blender.org/T36425#19 .
The cause of the issue was identified as roudning errors in Operators::createStroke() due
to insufficient numerical precision. Precision promotion from float to double was done in
the return values of getPoint3D/2D methods in Interface0D and its subclasses in C++
(data members stored in the 0D classes have already been in double precision).
before entering the recursion check.
Now use group tags instead of object tags, which could be a little more
efficient and was used before this patch too.
It still warns the user that there may be an error, but the baking goes
on. Also using the new is_uniform_scaled_m4() instead of float comparison.
Reported and fix suggested by Campbell Barton as a concern over 2bfc3deb
The remaining functions in blender_python.cpp changed from using the
MACRO to use python_thread_state_save/python_thread_state_restore
Since this bug only happens when 'Persistent Images' is on it was
introduced in some of the early merges with master and I never caught
it.
Thanks Daniel Salazar for helping with the bug hunting.
If we are using a mix node we still need to evaluate the BSDF lighting
even if scattering is successful.
Note: this was working for branched path (probably an oversight when
branched path support was introduced for baking, a good oversight though
;)
The scene file provided by the problem report has many degenerate faces coming from
a particle system. These zero-area faces were not expected in the ray-casting line visibility
algorithms of Freestyle. Now degenerate faces are properly excluded from the imported
mesh data and not fed to the line visibility algorithms.
The new recursion check for groups duplicating themselves has to un-set the LIB_DOIT flag after each object, otherwise it will prevent duplicating a (non-recursive) group multiple times.
To get the versioning code properly work, the condition should have been:
DNA_struct_elem_find(fd->filesdna, "FreestyleLineStyle", "MTex", "*mtex[18]")
The present commit uses another new structure member instead, to avoid referring
to the magic number 18 in the last string literal.
This was because to the filtering code, those FCurves still weren't in any
groups, and so couldn't be visible (since a temporary group is created to
house them). As a result, the visible-channels list would be empty, causing
all hidden FCurves to be treated as hidden.
These were broken by 1f3655d224, since
an argument of the wrong type was getting passed to ANIM_animdata_filter(),
resulting in no channels ever being picked up for the "visible channels" list.
Turned out there was still quite a few cases were indices were set dirty,
but elem_index_dirty was not tagged accordingly (mostly for BM_LOOP,
but a few others as well). So probably this crash was not the only one
hidden here.
Hopefully all possible cases were catched this time!
This fixes most of the cases, the only situation not addressed is when
the highpoly object(s) has non-uniform scale.
mul_transposed_mat3_m4_v3() should take care of non-uniform scales so
I'm a bit confused on why it doesn't work. The lowpoly object can have
any transformation, the only issue is if the highpoly object has
non-uniform scale.
Test file of the remaining issue:
https://developer.blender.org/file/info/PHID-FILE-tpw2xgddyzxtpg3e7xzs/
Reference reading:
http://www.unknownroad.com/rtfm/graphics/rt_normals.html
bleeding at the symmetry edges.
We need an extra way to detect if faces overlap here. An easy way is to
detect the winding of the faces in UV space. If the winding differs, the
faces will naturally overlap. I have tried a few approaches here such as
choosing an offset point from the middle of the edge for intersection in
both faces of the edge in uv space, but winding is the safest way and
should work with very small faces/dense meshes too.
Check that up_axis is not aligned with bone was wrong in at least two aspects
(not working against negative alignement case, and since ages it seems,
using Z axis when bones are along Y axis...).
Also optimized a bit here, better to have a normalized version of vec_roll_to_mat3(),
since it needs normalized vector anyway, and we have to normalize it for the tests
before calling it anyway (so now, we only do that twice in Transform code, instead
of three times).
And we can perform aling test *before* calling vec_roll_to_mat3!
This information only really seems to be for helping developers track
down bugs, and isn't really something the user needs. So, lets not spam
their console.
The bug was caused by using negative numbers as the end for playing forever (or until the end of the sound is reached) in the library. This was used with speaker objects which have an end of FLT_MAX now instead and the negative number interpretation was removed. I hope this doesn't break anything else.
Revert small part of own rB8714ae09f894, which changed scale setting from absolute to relative
(was good in absolute, but bad because it breaks existing rigs).
Fixes T40027. This means we get more CPU usage again when using multiple CUDA,
but the impact on performance is too big a problem with the current code.
It can't be undoe with Ctrl-Z, but Alt-H works just fine.
So instead of breaking someone's workflow let's just add
an info about how to bring bones back.
So! First, frame for absolute shape keys: never allow a new key to have the same pos as an
existing one (this does not make sense). This way, the two workflows are possible (create
all keys and then animate ctime, or animate ctime and then create keys where you need them).
Also, fixed UIList for shapekeys, the "absolute" test was wrong, and better to show frame
value, even though not editable, than nothing in case of absolute keys.
And finally, add getter to RNA 'frame' readonly value, so that we output real frame values,
and not dummy internal ones (which are /100) in our API.
On windows "." was replaced with the root directory.
On other systems any path starting with a '.' would be replaced with "/"
This was added for the file selector only,
so better handle this in the file selector (though it looks not to be needed).
rB568f0c7 added a recursion check that is supposed to prevent cyclic
cases where a group includes itself via dupli instancing.
The check function was descending into all groups nested inside the
target group - which works for single level recursion like in the test
case, but does not handle generic recursion. Basically it asked:
"is object X in the group already or in any instanced dupligroup?"
The new check instead asks:
"is group G dupli'd by X or any instanced subgroup thereof?"
which is what we really need to know.
The reported Freestyle warnings were due to wrong normals of filled faces at both ends
of a 2D extruded curve. The problem is detailed in the comment #19 of T39669.
The cause of the bug was an inconsistency in the use of vertex indices between
BKE_mesh_nurbs_displist_to_mdata() and init_render_curve() in the case of
DispList::type equal to DL_INDEX3.
This commit also fixes a related bug that the normals of filled faces were not inverted
when a scale of the curve object is set to a negative value (e.g., the Z scale was -1).
Reviewers: campbellbarton
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D532
The Bullet 2.82 update uses a different method for ray casting that
seems incompatible with our older files. So, for now we just force the
vehicle physics to use the older ray casting method.
This updates the fix in rB27db75363, which had to be undone
because it broke other bevels.
It also fixes cases where edges went away went doing vertex
bevel on vertices with some wire edges.
Logic here was very stupid, texboxes would have their initial
locations scaled by font size but not their width/height.
Now its possible to change font size while keeping the textbox layout.
Other fixes
- tab character didn't work properly with textboxes.
- memory leak when VFontData was missing.
For now only disables scene render if all the layers are disabled, but scene
will still be rendered if it's used in compo, has enabled layers which are
not used by compo.
Current pipeline doesn't allow to handle such cases nicely, so leaving it
for later.
It was a regression since 5d49eff. Not really sure about proper solution
here, so used a bit workaround-ish way for now.
Hopefully new cloth will be landed after this GSoC anyway.
- underline faces had flipped winding by default.
- BKE_vfont_to_curve_ex disallowed 0 underline height (annoying when sliding value)
- disallow negative underline height since it flips underline direction (just change position + height)
Problem happens when removing a rigid body reference in a constraint,
and then jumping to the start frame right away. This will cause a full
rebuild of the rigid body world. However, the btRigidBodys are removed
before the constraints, and this leaves dangling pointers in the
btTypedConstraints, which causes corruption when deleting those
constraints later.
Fix for now is to explicitly delete constraints in advance when
rebuilding, while they still have valid btRigidBody pointers.
Ultimately the whole memory management and ownership of Bullet data
needs redesign. This is already happening in the particles_refactor
branch and could be ported to master separately:
https://developer.blender.org/diffusion/B/browse/particles_refactor/source/blender/blenkernel/intern/rigidbody.c
ANIM_editkeyframes_refresh was testing handle selection as if those handles were transformed.
This is already handled by areas which need it,
so simply replace testhandles_fcurve -> calchandles_fcurve.
This was causing other bugs such as inserting a keyframe changing handles of unrelated fcurves.
Issue here is that upon entering sculpt mode, the mesh (and the object
mode) is stored in global undo. Now made the code similar to edit mode,
but since we don't really have any operator to push, this is just
ignored for now.
I have tried just disabling the sculpt toggle operator undo flag but
this didn't work due to the nature recursive of the operator calls
Undoing nodes that do not belong to the current object will cause the
saved bmesh log entry to be reverted instead. This entry can belong to
another object though.
This is easy to fix by enforcing name matching (this was borrowed by
edit mode but can definitely be improved) between current object name
and undo node name and deleting older entries.
However there are complications. Deleting dyntopo entries in this way
can leave a brush stroke as first dyntopo log entry. This can present
issues if we attempt to delete that entry since it's deleted mesh
elements may now have had their ids (which would still be valid at the
time) cleaned up. This can result in crashing if we attempt to resculpt
on the mesh. To fix this I have disabled releasing the deleted entries.
This entanglement between bm_log and undo is quite volatile but I hope
the system works better now.
Also minor cleanup, fix unneeded check warning
For some reason SCons defines _DEBUG, not DEBUG as mathutils was
expecting it to be.
Made it so mathutils checks for NDEBUG which mimics BLI_assert
define.
Own regression since e08db08. CustomData_copy_data() would fail in cases
when mvert/medge/mloop/mpoly arrays were lazy allocated since that change.
Now made it so this layers are copying from own copy of the arrays.
Not sure if this still misses some CD to be copied, from quick glance
seems no, but some further testing wouldn't hurt at all.
We need to support cutting degenerate ngons, see: T39418
This commit disallows cuts across faces where the same vertices can create better cuts on different faces.
as duplicators.
This property was always hidden in the UI for empties and cameras. It
doesn't make sense for the objects themselves (they are wires-only), but
also gets inherited by duplis. Now show it greyed out if not used, but
make it available for duplicators.
The shadow render passes could set a mesh's modified status to false
even if they were not rendered. This means their display lists do not
get updated. For now, just skip setting all buckets' modified to false
during shadow render passes.
This was suggested by @zanqdo on IRC. Hiding sockets is a user choice
(not to be confused with "unavailable" disabled sockets). Hidden sockets
suddenly popping up when linking is confusing and intransparent, better
just ignore them for the swapping.
with correct transform for Blender Internal.
According to previous code the obmat must be modified even if the
duplicated object itself is filtered later. TBH i have no idea how/why
this works, but nobody else does either ... All obmats are restored
after BI messes with them during render, so should be fine ...
Instead of 95, we can use 145 images now. This only affects Kepler and above (sm30, sm_35 and sm_50).
This can be increased further if needed, but let's first test if this does not come with a performance impact.
Originally developed during my GSoC 2013.
The issue was that we can't assume we support the colorspace from the file. The reported file had an invalid colorspace in fact, which was leading to the segfault in Blender.
Thanks for Sergey Sharybin for the help here.
airbrush + small brush size.
Fast textured strokes meant that sometimes the update width (calculated
from stroke distance) would be greater than the new ibuf width. This
meant out of bounds access for the new ibuf.
It's totally strange that I couldn't reproduce this issue on older
versions since the logic here has not really changed much between
versions.
This reverts commit 97823f6047.
I was to push a fix based on Brecht's solution (this commit) when he did
the same.
This fix is correct, but it misses replacing WM_JOB_TYPE_OBJECT_BAKE
with WM_JOB_OBJECT_BAKE_TEXTURE in the rest of the file, which may lead
to problems when calling the operator in a quickly sucession -
WM_jobs_test also tests for the same type of JOB.
I created WM_JOB_TYPE_OBJECT_BAKE elsewhere, may as well use it. Unless we revert 20c90ea and f194da3.
I'm fine with either way, just trying to get master to work again ;)
bNodeTree blocks.
This was broken by rB6e99fb0 (own commit). I expected the `do_action`
argument to be of no importance in this case due to node trees using
material animation, but this is not the case.
Anyway, this patch adds back a do_action to the BKE_libblock_copy_nolib
function as well to restore the previous behavior.
Although these crashes were quite sporadic, they seemed to happen most when
rmb-dragging strips randomly in quick succession. The most likely cause seems
to be a null check I accidentally took out during one of my commits yesterday.
Dashed lines were resetting stroke segment visibility by mistake (making invisible segments
visible, for example when applied to the results of square blueprint geometry modifier).
Frame nodes still have the "hidden" flag like all other nodes, but this
has to be ignored during resizing. width/height range for the frame
nodes must be unlimited for this to work correctly.
This makes it easier for finding the active or first selected channel
(where actual data channels need to be favoured over expanders - which come
first), as previously, long switch statements were needed everytime.
Previously mesh objects were created for individual visible stroke segments (recalling that
a stroke may consist of alternate visible and invisible segments). This could result in an
excessive memory consumption in stroke rendering when strokes are composed of many
short visible segments (for instance by means of dashed lines and/or blueprint geometry
modifiers).
Now mesh objects represent individual strokes instead of stroke segments, which tends to
reduce the number of generated mesh objects by an order of magnitude.
Code with contribution from Brecht Van Lommel (he wrote the
bake_differentials actually ;)
Reviewers: brecht
Differential Revision: https://developer.blender.org/D510
Show that modifier may not be supported in dyntopo rather than sculpt
mode in general. This is useful information since users may confuse the
two. Multires for instance is surely supported in sculpt mode, but not
in dyntopo.
Warn when a modifier does not support optimal drawing and mask/hidden
parts display.
User report was that hidden parts became visible after adding a new
modifier subdivision. Subdivision was added with the Ctrl+number keys.
First part of the fix was to make sure we add a multires instead of the
subsurf modifier in sculpt mode with the PageUp/Down Ctrl+number keys.
We can't really stop users from having added a subsurf modifier already.
We could be "smart" and convert subsurf to mulrires upon entering sculpt
mode maybe but that can easily backfire - or users may actually want
that. For now, just warn that the modifier won't support Hiding/Masking
and optimal drawing.
Multires did not support hiding in zero level. It's a case of PBVH not
being used in drawing due to code reporting active modifiers being used.
However multires level zero can be treated as effectively no modifier
and we can use mesh PBVH.
Now that Freestyle employs a separate Main, this workaround is no longer necessary.
Reviewers: sergey
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D513
It seems that since the changes to how new NLA strips are handled, accumulating
these in a temporary buffer and then trying to combine the results didn't
work that great anymore.
Not that happy with this, but this Item panel is a bit stupid, and we could add more
useful settings as well. So better get a design decision first, here.
This uses a different operator than the other time editors (as it needs to support
the setting of the value-cursor too), so the changes here didn't get propagated through.
This patch removes the limitations on the number of action layers in the BGE.
BL_ActionManager currently uses a fixed array to keep track of the action layers. This patch replaces the fixed array with a map which allows for dynamic allocation of action layers. Layers (map items) are automatically removed on BL_ActionManager's update function. The maximum number of layers is roughly the value of a short. Backwards functionality is maintained and there are no changes to the Python API.
Task Discussion:
https://developer.blender.org/T39572
Author: Kevin Ednalino
Reviewers: moguri
Differential Revision: https://developer.blender.org/D491
This fixes the SSS Direct/Indirect passes as well as the Combined pass.
Patch reviewed and with fixes and contributions from Brecht van Lommel.
Note: displacement/bump map (related to the report) will be handled separately
Reviewers: brecht
Differential Revision: https://developer.blender.org/D503
Now the COMBINED pass includes the Ambient Occlusion.
This was not reported anywhere, but while working in the Subsurface Scattering I realize we needed this fix for combined.
data.
Saving a file with a new blender node that uses bNode->storage data and
then loading that in an older version will make the node undefined, but
still retain the original type identifier (in case it is defined later).
If the file is then saved over and loaded again in the newer version,
where the node type is defined, it won't have a valid storage struct.
To handle such cases gracefully, check if storage data is expected but
doesn't exist when initializing node types. User then at least get a
chance of fixing the problem manually.
Suggested fix by @brecht.
As reported by ZanQdo on IRC, the bend modifier's "angle" property had a range
of +/- 572.95 degrees (equivalent to 10 radians). From the looks of things, this
was simply something overlooked when doing degrees to radians conversions. Now
it has a range of +/- 180 degrees.
Legacy texture shaders from the original stand-alone Freestyle program are also
declared as deprecated, in favor of Blender's new line style textures.
Patch contribution by Paolo Acampora. Thanks!
Just a case of uninitialized material array.
Make sure object materials are initialized properly when drawing UVs.
This might look strange since we do not really need materials here.
However, the same GPU object may be used in a 3D viewport and if we try
to avoid initializing the materials somehow this will break.
Also, this breaks on full screen UV editor as is apparent from this
report.
All image colorspaces should be displaying correctly in the UV/Image Editor.
Viewport may still present an issue, but it was reported separately
anyways (T40055)
Patch reviewed (and corrected) by Sergey Sharybin
Sculpt mode drawing fails after deleting a subsurf modifier
in sculpt mode and undoing.
This was quite difficult to spot. Main cause was that mesh data was not
synchronized properly between undo and sculpt code because we generated
a pbvh on derivedmesh invalidation without really refreshing the rest of
the data. This could result in undo and drawing operating on different
data.
To solve this and avoid bad level calls I had to move quite some code
around. Crazyspace is now moved to blenkernel, as did some sculpt calls
that make sure sculpt data are properly refreshed.
The previous fix was not quite correct:
* It would use the same wireframe/bbox check in viewport- and regular render
* The duplicator loop in Blender Internal is more permissive now to avoid disabling wireframe duplicators, but this means we need to check is_object_hidden right before making render objects to ensure this still works.
It's still gives some slowdown when painting a locked
key in the solid view, but since shape key is now longer
being re-used by DM.
but this change should still give some degree of speedup
propagating delta onto the keyblock if i remember the
code correct.
This was faster for my AMD system but slower for Intel.
However with gcc4.9,-O3 I was able to get roughly the same speed before/after.
Revert since this isnt giving such clear benefits on most systems.
next time you're deprecating some platform, compiler, feature just either
mail in the ML telling about this or if you're not that lazy poke developers
online whether then can spot issues quick.
Or be ready to pick up stuff being done by the developers your change gives
headache to.
Avoid filling up buffers when total buffer triangles are zero.
Better still would be to tag a node as hidden when doing recreation of
the PBVH tree by checking for any visible elements. Original bug report
probably has to do with OpenGL doing something funky but hidden nodes
should be tagged as hidden to completely avoid iterating for painting.
This is to be done in a later commit.
Also some naming cleanup for consistency, GPU_build_pbvh_mesh_buffers to
GPU_build_mesh_pbvh_buffers.
from particle edit mode.
Check for object particle edit mode, not psys->edit. This won't get
freed on exit (which might be worth considering doing, maybe?)
* If channel_index = -1, the pushdown operator will try to use the active AnimData
block. To see which one this will be, check the properties (NKEY) region in NLA Editor.
* Added more detailed error handling for the various cases where things are not
in a valid state for pushing down an action. This is notably to stop scripters
from putting the NLA into an inconsistent state.
This now works (barring one or two minor glitches and the operator on the pushdown
button being a bit of a hack). The old drawing code is still in place though, and
will be kept around for a little while yet while the last kinks are ironed out.
This commit lays some of the groundwork needed to port the last of the NLA Editor
channel types that's still defined using custom code. By moving this code over
to the newer standardised framework, this will enable widgets to have tooltips
to make it easier for users to find out what various buttons do.
Currently this isn't hooked up to anything though.
Updating object IPOs is not currently thread-safe since it also updates
children. This leads to problems when parents and children are both
animated. For now, updating object IPOs is done in its own loop to avoid
threading issues.
Instead of displaying the "tweaking action" status indicator icon on the right hand
side jammed against the mapping/no-mapping toggle (pin), this is now displayed as in
place of the action icon used on the left (as the type indicator). This is less messy
than the previous situation, which was needed partly due to the abuse of an unsuited
icon.
1) The "pushdown" operation is the actual name for the functionality
previously represented by the snowflake. That is, pushing the active
action onto the NLA stack as a new strip.
The icon that is introduced here is the icon I originally wanted to
use here for this purpose (but couldn't at the time the NLA was coded
as we didn't have the master SVG available yet). I'm making this change
now to hopefully make the "animation-layers" intentions behind the NLA
design a lot clearer than they have been so far.
2) Also added a dedicated icon for representing that the active action
is in the "tweaking" state (i.e. we're editing a particular strip's
action). Previously we were abusing an icon designed for an entirely
different purpose, mainly since its identifier looked like it would work.
* The automatically calculated influence was overriding the value we were passing
into the dummy strip, making this seem like it doesn't work
* Made some tweaks to prevent some potential memory-related bugs
- nlastrips_ctime_get_strip() actually saves off references to the list of strips
it gets, so declaring dummy_strip in an inner scope like this could potentially
be quite dangerous
- Prevented a potential memory leak for the early abort when there are no strips
for whatever reason (it shouldn't occur though)
The issue has been introduced in 0985bb4, made it so curves and masks
stick to old behavior.
FCurves are still flaky, namely, grabbing both handles selected would
change the length of one of the handles which is not nice imo.
The issue was caused by the temporary CD layers being allocated for subsurf
meshes, same as we've got back in 881fb43.
In the long run this temporary storage is to be re-considered, but it'll also
imply re-considering of the Derivedmesh interaction as well. For now let's
use a simpler solution which is forbidding modifiers to call getArray for other
objects' derivedMeshes but use an API calls which would allocate local copy of
the data preventing race condition of shared data in DM.
The rendered preview didn't show active layer or wasn't updated if the
layer was disabled. Now the viewport always shows an active layer regardless
of whether it's enabled or disabled.
Expand Cycles to use the new baking API in Blender.
It works on the selected object, and the panel can be accessed in the Render panel (similar to where it is for the Blender Internal).
It bakes for the active texture of each material of the object. The active texture is currently defined as the active Image Texture node present in the material nodetree. If you don't want the baking to override an existent material, make sure the active Image Texture node is not connected to the nodetree. The active texture is also the texture shown in the viewport in the rendered mode.
Remember to save your images after the baking is complete.
Note: Bake currently only works in the CPU
Note: This is not supported by Cycles standalone because a lot of the work is done in Blender as part of the operator only, not the engine (Cycles).
Documentation:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Bake
Supported Passes:
-----------------
Data Passes
* Normal
* UV
* Diffuse/Glossy/Transmission/Subsurface/Emit Color
Light Passes
* AO
* Combined
* Shadow
* Diffuse/Glossy/Transmission/Subsurface/Emit Direct/Indirect
* Environment
Review: D421
Reviewed by: Campbell Barton, Brecht van Lommel, Sergey Sharybin, Thomas Dinge
Original design by Brecht van Lommel.
The entire commit history can be found on the branch: bake-cycles
New operator that can calls a bake function to the current render engine when available. This commit provides no feature for the users, but allows external engines to be accessed by the operator and be integrated with the baking api.
The API itself is simple. Blender sends a populated array of BakePixels to the renderer, and gets back an array of floats with the result.
The Blender Internal (and multires) system is still running independent, but we eventually will pipe it through the API as well. Cycles baking will come next as a separated commit
Python Operator:
----------------
The operator can be called with some arguments, or a user interface can be created for it. In that case the arguments can be ommited and the interface can expose the settings from bpy.context.scene.render.bake
bpy.ops.object.bake(type='COMBINED', filepath="", width=512, height=512, margin=16, use_selected_to_active=False, cage_extrusion=0, cage="", normal_space='TANGENT', normal_r='POS_X', normal_g='POS_Y', normal_b='POS_Z', save_mode='INTERNAL', use_clear=False, use_split_materials=False, use_automatic_name=False)
Note: external save mode is currently disabled.
Supported Features:
------------------
* Margin - Baked result is extended this many pixels beyond the border of each UV "island," to soften seams in the texture.
* Selected to Active - bake shading on the surface of selected object to the active object. The rays are cast from the lowpoly object inwards towards the highpoly object. If the highpoly object is not entirely involved by the lowpoly object, you can tweak the rays start point with Cage Extrusion. For even more control of the cage you can use a Cage object.
* Cage Extrusion - distance to use for the inward ray cast when using selected to active
* Custom Cage - object to use as cage (instead of the lowpoly object).
* Normal swizzle - change the axis that gets mapped to RGB
* Normal space - save as tangent or object normal spaces
Supported Passes:
-----------------
Any pass that is supported by Blender renderlayer system. Though it's up to the external engine to provide a valid enum with its supported passes. Normal passes get a special treatment since we post-process them to converted and "swizzled"
Development Notes for External Engines:
---------------------------------------
(read them in bake_api.c)
* For a complete implementation example look at the Cycles Bake commit (next).
Review: D421
Reviewed by: Campbell Barton, Brecht van Lommel, Sergey Sharybin, Thomas Dinge
Normal map pipeline "consulting" by Andy Davies (metalliandy)
Original design by Brecht van Lommel.
The entire commit history can be found on the branch: bake-cycles
The mask make sure the conversion only happens in a few areas of the
buffer.
New Functions:
* IMB_buffer_byte_from_float_mask
* IMB_buffer_float_from_float_mask
The functions are an adaptation of their maskless counterparts without accepting different profiles for the input and output buffers.
Review: Sergey Sharybin
The issue was caused by the wrong scene used to acquire render result for.
Now made it so render pipeline reports currently rendering scene to the job
via special callback.
This also solves missing tile highlight issue when rendering multiple scenes
in the compositor.
does not work as expected for Viewport and Blender Internal.
As @sergey noticed, this was caused by rBfbf8595, which disabled
rendering for wireframe objects in the viewport. However, the same
function `is_object_hidden` is also used for skipping dupli objects then
which may themselves be renderable.
Now use two variants:
* is_object_restricted is the previous test for general viewport/render visibility (lets wireframe duplicators pass)
* is_object_hidden includes the wireframe check for actual render instances
It was broken in 7544961 and the proper way is to make sure proper obmat is being
copied to object before the draw.
Doing obmat copy in advance doesn't really work because object might appear multiple
times in the duplilist.
Probably will not be noticed in most scenes. This helps reduce noise when you
have multiple lamps with MIS enabled, at the cost of some performance, but from
testing some scenes this seems better.
I'm not sure why this function ever increased the object's refcount. Any
place in the code that calls KX_GameObject::GetParent() has to turn
around and call parent->Release(). Forgetting to call Release() was a
common cause of memory leaks (in fact, KX_SteeringActuator was probably
leaking). If the refcount needs to be increased, the calling code can
handle calling AddRef().
show diffuse option not correctly calculated on opening a file.
Make sure we call pbvh_show_diffuse_color_set after building the pbvh so
vertex data are available to it.
The was actually caused by the way how Cycles uses objects
layers. It's not possible to rely on the fact that layers
are flushed from Base to Object. It's only valid when rendering
active scene.
Now made it so layers are used from the base.
This also updates the configurations to build kernels for compute capability
5.0 cards, when using and older CUDA toolkit version this will be skipped.
Also includes tweaks to improve performance with this version:
* Increase max registers on sm_30, sm_35 and sm_50
* No longer use texture storage on sm_30
This was more like a TODO than a bug, but wasn't difficult to support.
Need to backup image slots before doing re-compo in nodes editor.
If something breaks, lease poke me!
The .exe extension *is* needed on Windows for makesrna, contray to the
patch author's protestations otherwise. Omitting it breaks compilation
on mingw at least (i.e. "rename()" fails). Thanks to "PerfectionCat"
for narrowing down this bug.
makesdna on the other hand still seems to work ok from a quick test I did.
Unless more issues crop up, just reverting this part of the patch should
be ok, though IMO these changes were unnecessary in the first place.
Weights do not need perspective correction since they are calculated on
flat triangle coordinates. Only pixel coordinates, reprojected from the
weights need this.
This was already fixed for clone brushes but in
soc-2013-paint where we have stencil preview, it is apparent that the
stencil suffered from the perspective distortion issue too.
The formula was not consistent across Blender and behaved strangely, now it is
a simple linear blend between color1 and min(color1, color2).
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D489
Usual dupli object issue, sometimes it's needed that all the object in
dupli group have modified obmat.
Made it an utility function now, which is used by convertblender and
dupli draw code now.
Freestyle relies on render layer name (RenderLayer::name) to find the render layer that
corresponds to a scene render layer being rendered.
When the active render layer is disabled during preview rendering, the function
render_result_new() populates the list of render layers (RenderResult::layers) with an
unnamed render layer while setting the active scene render layer index to zero.
This commit fixes the missing initialization of the render layer name by referring to that of
the first in the list of scene render layers.
neighboring sockets when connecting to an occupied input.
This now works based on socket names rather than types, which helps to
avoid unwanted connections to arbitrary sockets. Sockets are considered
swappable when their names match, based on an alphabetic prefix followed
by non-alphabetic suffix.
For example:
Image - Image : matches
Color1 - Color2 : matches (used e.g. in cycles mix nodes)
Roughness - Rotation : does not match (suffix is still alphabetic)
The test for wire edges when reattaching was wrong, because
some newly made edges are wire at the point of the test.
This made some duplicate edges.
Need to track the original wire edges a different way.
This option (alongside the Ease In/Out/InOut options already available) aims to make it
easier to get an initial curve that looks closer to the one you were expecting, by
automatically picking whether Ease In or Ease Out should be used based on the type of
interpolation being used for the curve segment in question.
Notes:
* The types chosen may need some adjustments (e.g. using ease in-out instead of just ease in)
* This does break compatability with files saved in previous dev builds, but only
if you were using Bounce/Elastic/Back with "Ease In"
This operator is used to make sure that if/when you have multiple strips
using the same action, if you select these and run this operator, each
strip will be given its own copy of the action. This is useful if you
decide later that you want to start using an existing action as a base.
NOTE: This does not recursively go inside meta's, so care is still advised
in that case.
This commit changes the default strip duplication behaviour (Shift-D) so that it will
create a copy of whatever action it uses. Previously, it was very hard, if not impossible
in some cases to create a new copy of an action to start working on in the NLA.
If you want the old behaviour, you'll need to use ALt-D (Linked Duplicates).
(Note: Although the new Shift-D may not be so optimal in all cases, I've decided to go
with this version since it aligns better with the way this works for objects. Doing the
opposite for NLA would just have added to the confusion)
Now code explicity excludes wire edges from beveling
and reattaches the wire edges to one of the newly
created vertices after beveling.
Also fixes a bug where vertex-beveling a wire-edge-only
vertex would not reattach the wire edges.
So that users opening a .blend saved in 2.70.5 and above in an older version
get warning the file might not be 100% compatible (should have been done
already for 2.70, actually).
Edit existing animsys_refactor module to make able to execute more complex conversions
('to' can now be a callback, instead of a simple prop name), and add a new
Update Animated Transform Constraints operator that uses it to handle complex
conversion for this constraint (drived or animated properties).
Note this operator has to be called manually (from 'space' menu), will make this clear
in release notes.
Also, similar changes made in 2.70 are *not* addressed by this script (would rather wrote
new operators as/if need arise, but Transform constraint looks much more sensible that the others).
This op should not remain in more than two or three releases anyway, imho.
Editing the y-coordinate of the right handle of a keyframe via the Active Keyframe
panel in the Graph Editor, while both handles are selected and are both of type "aligned"
resulted in weird behaviours such as the x-coordinate of the right handle changing
(and rapidly starting to overshoot) but nothing else. However, this problem
doesn't occur when only a single handle is selected.
It turns out that the "order of computation" logic in calchandleNurb_intern() gets
confused in the case of both handles being selected, and results in a sub-optimal
handling of the right handle being the one that's been changed. We hack around this
here by temporarily making it so that just the right handle is selected when doing
the updates here.
With the right handle selected, the movement of the left handle appears constrained
to the frame it is currently on, leading to unpredictable and wild overshoots of the
bezier curve. There appears to be little benefit in doing so.
The effect of this patch is that makes it so that instead of trying (initially) to
maintain the same distance between the two handles and then overshooting randomly
later, the handles now try to keep the same distance from each other (i.e. similar
doing a rotation around the keyframe) at all times. While this means that it isn't
possible to set up assymetric handles (i.e. where ease in to the key is less than the
ease out for example) using aligned handles (it's still possible using free; it's just
a lot more work to keep them aligned), the benefits of removing of the random blips and jumps
when things jump outweight the losses.
Patch by Brecht
This patch adds icons for each of the keyframe interpolation types (including
the easing equations), as well as icons for the easing type options.
Icons made by: Paulo José Oliveira Amaro (pauloup)
Reviewed by: Joshua Leung, Thomas Beck
This adds some view ratios in the video sequencer menu, based (copied) on the UV/Image Editor. It also fixes the inverted ratio issue reported in the same task.
Reviewers: #video_sequencer, #user_interface, schlaile
Reviewed By: schlaile
CC: jta, dingto, sergey, schlaile
Differential Revision: https://developer.blender.org/D447
Task: https://developer.blender.org/T37960
Thanks a lot :)
Those only cover the current set of brushes, soc-paint brushes will be
commited on that branch
* Buffer icons are usually in straight space (since we load from pngs)
so use src_alpha in OpenGL for blending.
* Allow blending for preview icons. This will be useful for the next
commit...
Make RNAPointer props un-editable here, we simply cannot handle this.
Also correct previous commit, asking for autonaming for all items was a bit extreme,
this is only needed for enums!
Broken enums widgets was a sequel of rBe5e0888a8f02 (when we want auto-naming,
we have to pass NULL, not and empty string!).
Now remains the RNApointers issue...
Text field part. Issue with enums dropdowns remains a mystery currently.
As for pointer fields, afaict they have never worked here, though it should
not crash.
UI_EMBOSS are values, not bitflags (own fault, most likely)...
Note we should probably get rid of UI_EMBOSST, it is used nowhere in UI code (set
in one place only, used nowhere).
This change makes lighting in GLSL preview more accurate, though it still
doesn't support material's "Exclusive" option.
Technical note: Changes in view3d_draw.c are not essential, these avoid
preparing unused shadow buffers.
Reviewers: brecht
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D457
Made the timeline option to only show keyframes from selected channels/data be a
per-scene setting instead of the per-timeline option it was previously. This makes
it easier for animators working on rigs with multiple bones (especially during the
polishing phase), since now the timeline and jump to keyframe operators use the same
setting to decide which subset of keyframes they need to consider.
By default, this option is enabled by default.
TODO: Extend this to the keyframe status shading on the active object name in the 3D view?
Note it was already the case for many simulations (cloth, rigidbody, etc.), just doing this
systematically in BKE_ptcache_add() now, will avoid us some recurring bug reports.
Advanced users will know when they can/have to raise this parameter!
In fact, armature layers operators (set layers, and show all) were kind of messy and broken
in Edit mode (Select layers had two different operators for Pose and Edit modes, both
using the same funcs that could only work in Pose mode, Show All was supposed to be
used in both modes but again, its exec code could only work in Pose one).
Fixed that by:
* Using only one op for each task, for both modes (with adapted poll func).
* Replacing 'object from context' access by an helper that returns the right Armature
object in both Edit and Pose modes.
This issue is because of a somewhat "special" behavior in old code, which got lost during rB09874df:
There was a variant of the `relinkConnections` function which would leave the socket completely unconnected. This is not a valid state really (given that each unconnected input must otherwise connected to a constant `Set` type node), but was used as a way to distinguish connected alpha/depth sockets in composite and viewer output nodes.
https://developer.blender.org/diffusion/B/browse/master/source/blender/compositor/intern/COM_InputSocket.cpp;28a829893c702918afc5ac1945a06eaefa611594$69
After the large cleanup patch ({D309}) every socket is now automatically connected to a constant, such that `getInputSocketReader` will never return a NULL pointer. This breaks the previous test method, which needs to be replaced by more explicit flags. Luckily this was done only for very few output nodes (Composite, Viewer, Output-File). These now use the regular SetValueOperation default in case "use alpha" is disabled, but set this to an explicit 1.0 value instead of mapping to the node socket.
Since autosnap was being applied during the flushing step for various practical reasons,
it wasn't possible to restore keyframes and handles to their original values, as these
would always get overwritten before getting a chance to be written back to the keyframes.
Note this is only a workaround in fact, adding some precision in radians case.
Validating the field will still generate a precision loss (doing otherwise is
doable-ish, but likely to backfire and/or add too much complexity in an already
complex area).
Since the NLA was also using the standard transform mode by default, this meant that
no frame/second step-snapping was happening here (i.e. like for the Graph Editor).
The actual differences between doing true stepping vs nearest here are quite small
(and don't really make much sense with NLA Strips), so for now, it should be fine that
these are the same.
The previous behaviour for nearest second meant that transforming the strips often
caused their lengths to change (sometimes drastically), since strip lengths aren't
always uniformly x-seconds long. Now, it only snaps the start frame value, and adjusts
the end of the strip to follow.
This works well for most cases, apart from negatively scaling the strip, where it will
get "stuck" as a 0.001 frame long strip (and the viewport drawing will be a bit weird
during this time). Nevertheless, negative scaling of strips isn't something that's exactly
recommended.
This commit cleans up and fixes some problems related to how the auto-snapping
behaviour in the animation editors works, resolving the issues mentioned in T39819.
1) "Nearest Frame" no longer snaps to the nearest second when time is displayed in seconds.
Instead, there is now also a "Nearest Second" option, so that either can be used
as needed instead of only when a certain time display is used.
2) A similar change has been made for "Time Step" - This is now "Frame Step" and "Second Step"
respectively.
Notes:
* Removed the unneeded getAnimEdit_DrawTime()
* Time Step/Frame Step don't work for Graph Editor yet (and seem to not have worked at all)
* NLA Editor also seems to be showing some weirdness now. Will be checked on.
* Cancelling nearest-second snapping doesn't work nicely, due to another bug with GraphEdit transforms.
Making both keymap and menu values the same, and adding a (debug only) check in
IDP_EqualsProperties_ex() warning when comparing two floats with nearly the same value.
* The button icon for jumping to the Render Layers tab of the Properties context is
now RENDERLAYERS. Previous icon BUTS was too general.
* Thickness position options are grayed out if not applicable. These options are applied
only when plain chaining is used with the Same Object option enabled.
Sequencer preview was rendering all scenes with the context
resolution, which doesn't give proper result. Now made it so
sequencer scene preview uses scene's resolution.
After recently reading through Robert Penner's original descriptions of these
easing equations for some other work I've been doing, I realised that the
ordering I introduced earlier was slightly off (with regards to sine and circular
types). This commit recitifies these issues.
When the dopesheet was open, "keyframe edited" events from the graph editor
(i.e. fired whenever any properties on keyframes or FModifiers are changed)
would trigger the dopesheet to synchronise selection states of anim channels
and ensure that FCurve autocolours are initialised correctly.
This however was undesired when editing properties in the graph editor. Now,
made it so that keyframe adding/removing operators use different notifier flags
to specify that the channels might have changed + need colour syncing, and
adjusted the dopesheet updating logic to fit
* Moving the BlenderDebugDraw (derived from btIDebugDraw) from
KX_BlenderSceneConverter to CcdPhysicsEnvironment
* Moving CcdPhysicsEnvironment initialization to CcdPhysicsEnvironment
(this could probably be cleaned up some more with some sort of
factory, or at least moving code to CcdPhysicsEnvironment's
constructor)
* Simplifying physics environment initialization (went from two
switches to one)
As suggested by Campbell, bump subversion and reset all ui_previews data (so that any file that could
have been corrupted in past two weeks because of this bug is fixed and valid again).
More like a workaround, current depsgraph does not really allow for animated layers,
esp. when we alwys expect at least one to be enabled (same issue as with basic Scene layers).
So, turned out after all we need a foreachmapped helper for loops as well... :/
CDDM and EBDM were reasonably simple, but I fought hours with CCGDM (subsurf) to find only a
bad working solution (see D478). So I fallback to a code similar to CDDM one.
Probably not nice for performances, but loops and subsurf are not matching well...
Reviewers: campbellbarton
CC: brecht
Differential Revision: https://developer.blender.org/D478
Also refactor:
- Material property UI related to shadows
- Preparation of OR-ed mode flags (ma->mode_l) of render materials
Reviewers: brecht
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D313
Problem was introduced back in 2.70 during Project Pampa when the FCurve Normalisation
feature was introduced. The cause was that the normalised cursor value was always getting
passed to the KeyframeEditData context, even when it wasn't needed.
These were used in BL_CreatePhysicsObjectNew() in the converter. However, all
of the data put into KX_ObjectProperties was then copied again in
KX_ConvertBulletObject(). So, instead KX_ConvertBulletObject() now gathers the
information it needs itself, which avoid this odd double conversion step for
physics.
As a side-effect, the old code would result in static non-mesh objects with no bounds
set to still have triangle mesh bounds. This would result in no bounds for these objects.
If a bounds was set that required a mesh, non-mesh objects would become sphere bounds.
This is now true regardless of whether user bounds were set. In other words, static
non-mesh objects now use sphere bounds by default instead of mesh bounds. This might
slightly alter some games, but these objects should generally be set to No Collision
anyways.
This caused a couple of fireflies in koro_final.blend. The wrong normal would
cause the shading point to be set as backfacing, which triggered another bug
with hair BSDFs on the backface of hair curves. That one is not fixed yet but
there's a comment in the code about it now.
This actually had nothing specific to new split normals, it was an internal limitation
of BI raytracer, which would check against neighbor face shadowing only when they shared
a common vertex, now it also performs checks when both faces have a vertex with a common
"ancestor" (org index).
Note this allows to also fix same issue when using SplitEdges modifier (and potentially
others?), but only when AutoSmooth is enabled (due to some compute/mem overhead, we
do not want to enable this code systematically).
Thanks to Brecht for advices and review!
Old algorithm:
Raytrace from one transparent surface to the next step by step. To minimize
overhead in cases where we don't need transparent shadows, we first trace a
regular shadow ray. We check if the hit primitive was potentially transparent,
and only in that case start marching. this gives extra ray cast for the cases
were we do want transparency.
New algorithm:
We trace a single ray. If it hits any opaque surface, or more than a given
number of transparent surfaces is hit, then we consider the geometry to be
entirely blocked. If not, all transparent surfaces will be recorded and we
will shade them one by one to determine how much light is blocked. This all
happens in one scene intersection function.
Recording all hits works well in some cases but may be slower in others. If
we have many semi-transparent hairs, one intersection may be faster because
you'd be reinteresecting the same hairs a lot with each step otherwise. If
however there is mostly binary transparency then we may be recording many
unnecessary intersections when one of the first surfaces blocks all light.
We found that this helps quite nicely in some scenes, on koro.blend this can
give a 50% reduction in render time, on the pabellon barcelona scene and a
forest scene with transparent leaves it was 30%. Some other files rendered
maybe 1% or 2% slower, but this seems a reasonable tradeoff.
Differential Revision: https://developer.blender.org/D473
Differential Revision: https://developer.blender.org/D298
Allows users on Windows to enter UNC paths in the filebrowser and to link to .blend files on a UNC path. Functionality is limited still, we can't browse the network yet and have no support to check user rights so far.
What works:
- enter an UNC path in the file browser manually or via copy/paste
- navigation within the UNC share subfolders
- link to a file on a UNC share
What does not (yet) work:
- browse the network for computers and shares
- browse to a folder that requires entering user credentials
Contributors:
Rob McKay - original patch
Campbell Barton - style fixes
Reviewers:
Campbell Barton, Brecht van Lommel
This was the original code to get things working on old GPUs, but now it is no
longer in use and various features in fact depend on this to work correctly to
the point that enabling this code is too buggy to be useful.
This can for example be useful if you want to manually terminate the path at
some point and use a color other than black.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D454
Pass the arrays by reference rather than by value,
should give some percent of speedup.
Also don't pass the dimensions to the function but
get them from the images themselves.
Hopefully this will give some %% of tracker speedup.
View2D had some inconsistencies making it error prone in some cases.
- Inconstant checking for NULL x/y args.
Disallow NULL args for x/y destination pointers, instead add:
- UI_view2d_region_to_view_x/y
- UI_view2d_view_to_region_x/y
- '_no_clip' suffix wasn't always used for non-clipping conversion,
switch it around and use a '_clip' suffix for all funcs that clip.
- UI_view2d_text_cache_add now clips before adding cache.
- '_clip' funcs return a bool to quickly check if its in the view.
- add conversion for rectangles, since this is a common task:
- UI_view2d_view_to_region_rcti
- UI_view2d_region_to_view_rctf
for one of the input shaders is zero.
This gives about 5% speedup for koro_final.blend. In general this is important
so you can design shaders that run faster for shadows, diffuse bounces, etc, for
example by skipping procedural textures or even using a single fixed color.
Toggle buttons are now placed to the left of those labels separating sets of line style
options, so as to save some space (in line with the new UI elements added in the
commit f60a66f417).
Line styles now have a set of new options for rearranging the stacking order of lines.
This gives artists more control to determine which lines should be drawn on top of others.
Two available sort keys are the distance from camera and curvilinear 2D length.
Since the distance of a line from camera may vary over vertices, another option called
integration type is used to compute the sort key for a line from the values computed at
individual vertices. Available integration types are MEAN, MIN, MAX, FIRST and LAST
(see the tool tips for more detail).
Seems like custom data indices can be mangled when many layers are
present in the mesh, so use more strict calculation of offset (copied
over from BM_CD_LAYER_ID).
I was under the impression that all custom data would be deleted on
entering dyntopo, it seems this is not the case though.
It makes code more tidy (avoids having to call invalidation on a myriad
places). Also makes sure other invalidation cases (some mesh change,
e.g.) work as expected.
This means if one makes change to either SConscript or CMakeLists.txt
there he MUST update bundle.sh.
Also made it so *.cc files from intern/libmv are matching which would
make it easier to add more .cc files there if needed.
And one more thing is removing 'simple_pipeline/distortion_models.cc'
which doesn't match any of the files.
Don't use a dedicated node layer but use temporary int layer instead.
Works like a charm as long as we are careful resetting the layer when
needed (after pbvh clearing and always after bmesh has been filled in
undo)
Tip by Campbell, thanks!
It shouldn't create issues in practice because modified face ids should
have been reclaimed from an added face on a previous entry, but add for
completeness.
This commit makes it so CameraIntrinsics is no longer hardcoded
to use the traditional polynomial radial distortion model. Currently
the distortion code has generic logic which is shared between
different distortion models, but had no other models until now.
This moves everything specific to the polynomial radial distortion
to a subclass PolynomialDistortionCameraIntrinsics(), and adds a
new division distortion model suitable for cameras such as the
GoPro which have much stronger distortion due to their fisheye lens.
This also cleans up the internal API of CameraIntrinsics to make
it easier to understand and reduces old C-style code.
New distortion model is available in the Lens panel of MCE.
- Polynomial is the old well-known model
- Division is the new one which s intended to deal better with huge
distortion.
Coefficients of this model works independent from each other
and for division model one probably want to have positive values
to have a barrel distortion.
Otherwise devices used for display will lock up the UI too much. This means
you might still get 100% CPU for the display device, but for others CPU usage
should be low still.
The check to see if a device is used for display may not be entirely reliable,
it checks if there is a watchdog timeout on the device, but I'm not entirely
sure that always exists for display devices or is disabled for non-display
devices, though some tools like cuda-gdb seem to make the same assumption.
Ref T39559
- pass label strlen since in many cases its already known.
- use single linked list for cached text drawing.
- add BLI_link_utils.h for single linked list macros.
own tests give approx 22% overall speedup.
(See commit e1771e72fbbf828dbf5bed871b814288389f3611 for more detail of
the problem).
Fix for a C++ scope issue caused by 'using namespace Freestyle' that was assumed
even within the Freestyle namespace definition.
(See commit e1771e72fbbf828dbf5bed871b814288389f3611 for more detail of
the problem).
Made changes to intern/view_map/Interface0D.h and intern/python/Director.h to
avoid #include <Python.h> and keep non-Python header files independent of it.
(See commit e1771e72fbbf828dbf5bed871b814288389f3611 for more detail of
the problem).
Fixed for #include <Python.h> not properly put in the extern "C" { ... } construct.
Also removed redundant inclusion of the header file in the Freestyle Python API code.
The issue was caused by including both Python.h and BLI_math.h. Since the macro
definition in Python.h was unconditional, it was necessary to include it first and then
BLI_math.h.
Addressing the issue turned out to affect many Freestyle source files. There were several
other coding flaws that complicated the situation:
1. intern/system/FreestyleConfig.h was abused just to include BLI_math.h which was in
fact unnecessary in FreestyleConfig.h. While addressing this, inclusion of both math.h
and BLI_math.h had to be duly dealt with to avoid a compiler error about round().
2. #include <Python.h> was not properly put in the extern "C" { ... } construct.
3. intern/view_map/Interface0D.h and intern/python/Director.h, both of which included
Python.h, were included from within Freestyle header files that were almost or completely
independent of Python. These non-Python header files were used everywhere in the
Freestyle code base, causing many cases of the reported macro redefinitions.
4. The Director.h file was also declaring 'using namespace Freestyle', which was causing
another scope issue in several Freestyle header files.
This commit is intended to address the first problem above. Subsequent commits will deal
with the other three.
This change introduces a new checkbox to deactivate the sensors, controllers and/or actuators. It is useful during the development phase to avoid delete sensors, controllers or actuators if you want to test something new.
NOC: The wiki page is being updated (the images mostly), but the feature is already in the 2.71 release log.
{F61628}
Reviewers: moguri, dfelinto, campbellbarton, dingto, #user_interface, billrey
Reviewed By: moguri
CC: billrey
Differential Revision: https://developer.blender.org/D16
This makes it easier to have per kernel number of registers. Also, all the
tunable parameters for this are now in kernel.cu, rather than spread over cmake,
scons and device_cuda.cpp.
Tessellated split normals CDlayer had no swap func, hence ugly little test_index_face()
could not rotate its values correctly when rotating tessellated faces...
Extra-reported in T39735.
There was some deadlock due to trying manage Python's GIL. Instead of
continuing to fight with it, anything needing to call into Python while
conversion during lib loading is just delayed until it can be done in
the main thread.
The problem here was that animation buffers got initialized with zeros in the beginning for unknown parts. Now it gets initialized with the first known value.
The bug's result was that the animation of the pitch started with 0 on first playback and thus any seeking while the pitch is zero resulted in seeking to the beginning.
For a long time, one of the bottlenecks when drawing summary channels in the dopesheet
(especially with many objects) was how the long keyframes feature (i.e showing holds
between keyframes) got built. Specifically, it was the step where we check on the previous
keyframe to see whether there's a hold between those two.
The old code performed some elaborate checks, which made sense back when we used to handle
certain summary channels (e.g. object-action/ipo, and groups IIRC) differently. However,
nowadays, everything just does it by going over the FCurves one by one, so the offending
code wasn't really providing much benefit. Unless I've forgotten some other reason why
that old method is necessary, this commit should provide a decent speedup here, making
things somewhat interactive now (if still a bit jerky).
Other Tweaks:
1) Introduced float-precision threshold when checking to see whether an existing long
keyframe could be reused. This should hopefully reduce the number of fp-jitter issues
when creating summaries for many channels, reducing the number of duplicates created.
2) Precompute colours used for shading the long keyframes, instead of recomputing for
each block.
Since this is now pretty much the de-facto "minimum distance between keyframes",
we might as well expose this in this header so that other places which need similar
thresholds can perform similar checks (needed for my next commit)
Issue fixed by:
* Not having constant width for all columns, but adapt each to its content's width;
* Adapting undo's menu height to undo list length (so that we never have more than three columns).
It is still possible to get issues in extreme cases (small screen, high DPI size,
long op names everywhere...), but this should now be rare corner cases.
Also fixes a minor glitch with undo menu (first column had one item less than the others...).
Many parts of the compositor are unnecessarily complicated. This patch
aims at reducing the complexity of writing nodes and making the code
more transparent.
== Separating Nodes and Operations ==
Currently these are both mixed in the same graph, even though they have
very different purposes and are used at distinct stages in the
compositing process. The patch introduces dedicated graph classes for
nodes and for operations.
This removes the need for a lot of special case checks (isOperation etc.)
and explicit type casts. It simplifies the code since it becomes clear
at every stage what type of node we are dealing with. The compiler can
use static typing to avoid common bugs from mixing up these types and
fewer runtime sanity checks are needed.
== Simplified Node Conversion ==
Converting nodes to operations was previously based on "relinking", i.e.
nodes would start with by mirroring links in the Blender DNA node trees,
then add operations and redirect these links to them. This was very hard
to follow in many cases and required a lot of attention to avoid invalid
states.
Now there is a helper class called the NodeConverter, which is passed to
nodes and implements a much simpler API for this process. Nodes can add
operations and explicit connections as before, but defining "external"
links to the inputs/outputs of the original node now uses mapping
instead of directly modifying link data. Input data (node graph) and
result (operations graph) are cleanly separated.
== Removed Redundant Data Structures ==
A few redundant data structures have been removed, notably the
SocketConnection. These are only needed temporarily during graph
construction. For executing the compositor operations it is perfectly
sufficient to store only the direct input link pointers. A common
pointer indirection is avoided this way (which might also give a little
performance improvement).
== Avoid virtual recursive functions ==
Recursive virtual functions are evil. They are very hard to follow
during debugging. At least in the parts this patch is concerned with
these functions have been replaced by a non-virtual recursive core
function (which might then call virtual non-recursive functions if
needed). See for example NodeOperationBuilder::group_operations.
We really should not have Object.imat_ren, that's the wrong place to put render
data. But just update it now on rotation rather than doing a bigger refactor to
remove it.
For now disable using linked edit mesh in the meshdeform modifier.
This is because editbmesh_get_derived_cage_and_final() might easily
conflict with the thread which evaluates object which is in the edit
mode for this mesh.
We'll support this case once granular dependency graph is landed.
Root of issue is probably in AUDASPACE itself (or even in OpenAL?).
Issue not present with SDL backend, only OpenAL afaik.
For now, workaround in Blender is to call seek twice when starting play
(not needed when resuming from pause).
Nice little mistake, since the invalid mem access only happened once (the first time),
was close to valid mem, and was only used to read, it would not crash often...
RNA is low level data access, setting mesh values shouldn't change draw options.
Also its redundant to name attrs Face.freestyle_face_mark,
removed edge/face from attr names
This revision also addresses the issue of these properties not shown in the Outliner
in the Datablocks display mode. Now Freestyle edge/face marks can be keyframed
through the Outliner.
- BLI_ghashutil_strhash_n takes string length, to avoid terminating the string before hashing.
- BLI_ghashutil_inthash/uinthash take ints, to avoid casting to (void *)
This also showed up incorrect use of inthash, which was using a pointer.
This is done in SCons and the only reason to disable OpenMP on
Windows was to deal with the express version of compilers.
Now when we are about to switch to msvc2013 by default we can
safely enable OpenMP there since even express version of this
beast supports OpenMP.
This was rather a TODO item related on supporting the proper painting color
space, but added a small tweaks which will make things working for now in
most of the situation (assuming the default view of display is ivnertible,
as it is to be expected to be anyway).
Shouldn't give much overhead since the conversion processors are cached in
the color management code.
And for the note: no, projection painting does not requite such a tweak
because viewport works in sRGB space anyway.
Either was some residue from the past or somebody didn't implement this in
the right way.
Also fixed memory leak in ED_space_clip_color_sample() caused by missing
image buffer release.
Naming:
Change pressure to size_pressure, it notes correctly that this value is
updated and expected to be used for size updating only.
Change name of cursor function and since it is used for uv sculpting
only now move to the relevant file. Also cleanup unneeded functionality
from function.
Stroke:
Separate updating of stroke variables to invariants (updated when
stroke->init = false) and variants.
Idea and code by Brecht, many thanks!
Reviewers: brecht
Reviewed By: brecht
CC: campbellbarton, dingto
Differential Revision: https://developer.blender.org/D369
Note that this commit completely replaces old behavior of the auto_smooth feature in BI.
Also note that split normals are only handled when no "advanced geometry post-processing" is used
(something like Displace will obviously break it, since it has to re-compute normals after displacement...).
Reviewers: brecht
Reviewed By: brecht
CC: campbellbarton
Differential Revision: https://developer.blender.org/D368
* Merely a re-implementation of core split algorithm for BMesh, taking advantage of topological data available.
* This code needs valid loop indices, so added BM_LOOP support to BM_mesh_elem_index_ensure() & co.
Reviewers: campbellbarton
Reviewed By: campbellbarton
CC: brecht
Differential Revision: https://developer.blender.org/D366
* Add a new calcLoopNormals function to DerivedMesh struct, and implement it for CDDM and CCGDM (subsurf).
EditDerivedBMesh (edit mode DM) only gets a dummy one in this commit.
* Add a tessellated version of CD_LOOPNORMAL layer (CD_TESSLOOPNORMAL), with relevant code to handle it
(tessellation, rna access, etc.).
* Change auto_smooth options of Mesh (angle now in radian internaly, and toggle is now used to enable/disable
split normals in DM creation process). Note BI render code is not touched here, hence its behavior regarding
this option is now incoherent, will be addressed in a separate commit.
Reviewers: campbellbarton
CC: brecht
Differential Revision: https://developer.blender.org/D365
de-duplicate object drawing code between offscreen buffer and regular view3d.
Also skip drawing grease pencil and setting pixelspace when drawing from lamps.
Fixed the part with missing tiles highlight and render info when rendering
different scene via the render layers node.
Displaying of the rendered result for a different scene after the render
is finished is still not "fixed". That's an intended behavior actually
to display render result for an active scene.
HSL sliders jumping. The issue here is that we store HSV in display
space.
To correctly account for that, made sure HSV/HSL is now using the
displayed
color to compute the values in the sliders. RGB values still show
the property values (linear for linear, gamma corrected for gamma
corrected)
We could change the way we store HSV uniformly (ie in all the code), but
we would need to add many more conversions in the picker code to account
for storing it that way. Also it doesn't make sense: Color pickers
should help with -visible- color selection. It may
be worth changing the RGB sliders as well.
My fix takes into account the way HSV is stored in HSVCube as well in
the code.
The issue was caused by the readEWA spending loads of time trying
to sample regions outside of the buffer.Solved by adding an early
exit check.
We could also clamp the sampling region to the rect, but it's
not so much clear whether weight will be correct in such case
so left it for the future.
Addition of the path to the Freestyle Python API modules to 'sys.path' was delayed until
the first Freestyle rendering, so that any import attempt of the modules in the Python
Console always failed. Now the update of 'sys.path' is done at Blender start-up.
This allows the Freestyle-specific modules to be imported without running Freestyle,
facilitating quick interactive testing in the Console.
These modifiers were not working properly when they were applied to strokes
whose backbone was already modified by other geometry shaders. This problem
was due to the use of Normal2DF0D that compute 2D vertex normals based on
the underlying FEdges up on which initial stroke geometry is defined. Now vertex
normals are computed on the basis of modified stroke vertices.
A helper function 'stroke_normal' for computing normals of stroke vertices was
added to the 'freestyle.utils' API module.
This gives a huge speedup gain for cases when you've got
rather huge markers on a byte images.
Done by skipping IMB_float_from_rect()/IMB_rect_from_float()
for such cases. We can sample the buffers without color space
conversion.
Our deformer system really needs some work. First, there was a crash
with shape keys because BL_ModifierDeformer derives from
BL_ShapeDeformer, which means we try to execute shape keys even if we do
not have them. Also, for some reason BL_ModifierDeformer::Update() does
not work if called from the threaded loop, so it is skipped for now. In
other words, skinned updates on meshes with modifiers are currently not
run in parallel.
This reverts commit 6cc5bdc99e.
Revent this for 2.70a, it changes behavior too much without allowing
keyframe handles to be scaled some alternative way.
This was a nasty bug which was caused by specific of how face-edge
attributes are stored in Carve.
Face pointer is used in the map key which works just fine in all
cases except for the cases when some face is getting freed after
it was stored in the map.
This might give real issues when new face is allocating because
it's possible new face would have the same address as the freed
one.
Such cases used to happen when union of separate manifolds is
needed for the operands AND jemalloc is enabled.
Solved by dropping attributes for the freed faces from the map.
Maybe not the fastest ever approach, but not sure how to make
it faster actually. Should work just fine. It only happens for
complex setups with intersecting manifolds in the operands.
The variables are considered invalid unless DM_update_materials is
called prior to use. Only use case currently is
mesh drawing. This helps with excessive allocation on the stack during
GPUObject creation, but may help elsewhere in the future as well.
-ffast-math is evil, not sure why it was enabled...
I seems to work better on OSX but it's still not a good idea.
The SConscript for bullet is a mess, I don't understand why
we use different flags for different platforms in the first place.
Seems to be a historical artifact but I don't know enough about scons
to try and clean it up.
This fixes the ptaxs "ACCESS_VIOLATION" error and should allow our Linux and Windows build bots to compile again.
Unfortunately this comes with a performance penalty on sm_2x cards, so this is only a workaround for now. Branched Path is still globally disabled on GPU.
Issue here was that buttons_texture_context_compute() was getting scene from (button-customized) context,
before the button paths (and hence, context) had been updated. So after an undo, it was getting an invalid
(freed by undo) scene pointer.
Now update BCONTEXT_SCENE path before calling buttons_texture_context_compute().
There is no good solution here, since RNA props can only have one type/unit.
Tried to find the less worse one - have different RNA props for same DNA value
(a bit like the angle/length for camera lens).
Also fixed two other issues with Transform conctraint:
* Angle were still in degrees (yes, another backward-compatibility breacking).
* Scale was absolute, unlike loc/rot.
Also cleaned up a bit the code, replaced some magic numbers by proper enums, ...
This required BL_ArmatureObject to have tighter control over armatures and poses.
Also, (Blender) armature objects are now copied instead of shared between
BL_ArmatureObjects to avoid race conditions. Also, due to the armature copy,
shape key drivers need a bit of extra fiddling to get the correct armature copy.
Initially OpenMP was used for threading, but then BLI_task was used due to being
less compiler dependent.
This commit also places time spent on skinning updates in the Animation
profiler category (was previously under the Rasterizer category).
Doubts: are written as comments in the code
Current Problems:
- stereo modes sometimes have different names and are ordered differently everywhere (blender properties panel names and tooltips, blenderplayer cmndline)
- dome mode as several options, but can only take one at a time, like -D mode bla -D tilt X -D ...
- gameoptions are also given one at a time, like above, but with an assignment -g option = X . This is inconsistent.
- gameoptions need to be reviewed and docummented
- parent to window (-i) is useful for what? (there is no wiki on this either)
- there is no support for verbose options like --help (blender itself supports it)
Reviewers: dfelinto, moguri
CC:
Differential Revision: https://developer.blender.org/D312
See T39470 and D436. Code by @tippisum, with some minor edits by @mont29.
Tested with various rigs, including Rigify, CGcookie flex rig, and gooseberry/pataz caterpillar.
Riggers, please test it, no change expected in behaviour.
Reviewers: aligorith
CC: tippisum
Differential Revision: https://developer.blender.org/D436
Issue is, looks like Debian testing's graphic backend now uses LLVM-3.4,
which conflicts with 3.3 in Blender and makes it crash.
So add support to handle 3.4, by both allowing to force a specific version for some libs,
and switching back to head of official repo for OSL.
These data elements are undocumented and of little use. For now they are commented out
in the implementation in favor of less memory consumption, and a very limited support for
these data components in the Python API was just removed (should be easy to recover).
Symbol 'real' is an alias of double and is subject to future change, while the interface of
0D/1D functions is part of the stable Freestyle Python API. So all occurrences of this type
in the class definitions were replaced with double.
Almost all pools allocated 2 chunks on initialization,
every element needed to be added to the free-list which
would never be used for small pools.
Now allocate only one, gives minor speedup for some bmesh operations.
This function used ugly hack with static variable which was
preventing some type checks in DAG nodes. Using this variable
form multiple threads is not considered safe, apparently.
Solved by moving this variable inside the DAGForest structure.
so it's global for the graph now, but different graphs does not
run into conflicts.
This required passing the forest to some functions, which doesn't
look so much nice, but don't want to spend time on making this
code look beautiful because it is really to be replaced by the
new dependency graph.
This is really bad bug actually which is must go to 'a'.
In practice this means that if you don't connect a texture to your volume nodes
it will figure that out and render the node faster, rather than you having to
specify it manually.
Main weakness is custom OSL nodes where we have to assume it is heterogeneous
because we don't know what kind of data the node accesses.
This actually implements the idea used in Gimp which is grabbing
an arbitrary point on the spline and dragging it, ensuring spline
goes over this point. This is really useful way to tweak spline
curvature.
Currently only affects on a closest handle, meaning no weighting
on changes for both handles which are adjacent to the same segment
will happen just yet,
Another limitation is that currently such a slide is a big jumpy
when you start sliding. This is because projection is not used
to calculate u value because projection used to fail a lot for
me here and didn't find a nice solution for this yet. But this is
to be improved for sure!
Vertices of two edges were swapped by mistake. Also fixed indentation and added
a couple of debug prints to make it easier to visualize the lines using Matlab.
Issue here is that "show diffuse" option does not respect its intended
purpose which is to be used only for masking.
There are a couple of caveats here:
Dyntopo and multires -always- have mask data enabled, and thus as soon
as one goes to dyntopo mode or adds a multires modifier he would get the
default grey color instead.
Matcaps would break when nodes asked for a diffuse material color (this
was broken before too). Solved by adding global material state for when
matcaps are enabled. Also matcaps don't always played well with VBOs
off.
Added a few more missing updates for mask operators to notify
show_diffuse property as changed. This was also needed on rebuilding
dyntopo pbvh.
Also make zero mask color duller again after artist feedback.
Was marked as a todo in the code.
This does not yet take care of correct display for multi material
meshes, since it would need correct separation of faces during pbvh
creation. Instead we just take material of first face in node and assume
that the rest faces have the same. This will create some funky effects
if one attempts to sculpt in this way.
Note: This does not yet address T39517
Those variables would get declared on fragment shader level and since we
use reserved opengl variables, some compilers would throw an error
(NVIDIA allows, some ATI compilers may break). Instead, use a separate
opengl built-in category especially for them. This works on NVIDIA, and
will wait for tests of this commit from ATI users.
This deduplicates/simplifies some code. Also cleanup up a bit scopes UI code!
Use new GRIP button for uiList grab-resize.
This allows us to greatly simplifies the code, and get rid of a few hacks in uiList event handling!
Note autosize mode of uiList is now trigered by any value of list_grip below a given threshold, rather than the fixed zero value...
Reviewers: brecht
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D343
This is done by adding a new button type, GRIP, similar to other numbuttons
(scroll, slider, ...), which here controls the preview height.
Then, we add a new DNA struct to be able to save that height in Blend files
(note I choose not to use Panel struct for this, because we would then have the
same limitation we used to have with uiLists, only one preview per panel
and no preview outside panel).
This implies a change to template_preview UI RNA/py API (each preview needs an ID),
but this is backward compatible, as by default datablock type will be used if no ID is
given (which means e.g. all material previews with no ID will have same height).
Reviewers: brecht
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D342
We need to salt temp exr filenames with pid, else several instances of Blender rendering
the same scene on same machine and using "Save Buffers" option would use the same files!
There is not much sense to have a whole BLI file just to check SSE2 on CPUs...
So idea is to rename it to more generic "BLI_system", and add to it more system-related
utils, like e.g. an include helper for getpid(), which allows to hide unix/windows
internals from rest of the code...
Reviewers: campbellbarton
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D439
render layer nodes in a pinned tree from different scene.
The way these updates work is a nasty legacy hack:
https://developer.blender.org/diffusion/B/browse/master/source/blender/nodes/composite/node_composite_tree.c$277
This function is called //very frequently// by the get_from_context
method. However, this does not get called for pinned node trees, so
when showing a different scene's compositing nodes in the editor they
may not get updated correctly.
Now moved this update call out of get_from_context so it happens in any
case. Will be called no more frequently than before (on every redraw).
Eventually the depsgraph should handle this more precisely, it's just a
simple ID dependency anyway ...
triangles.
It is possible to end up with such nodes using brushes in aggressive
collapse mode. Those nodes should normally be cleaned up, since they can
never be actually reused (adding more geometry to a node requires the
node having some geometry to begin with) but until we support dynamic
nodes, better delete those to avoid binding graphics driver resources.
If such zero elements buffers were used, GL error out of memory would be
reported.
When node face gets deleted, added or exchanged, the nodes should update
their draw buffers, normals and bounding boxes. This was not being done
before so there were graphical glitches apparent, especially in collapse
mode.
This commit does various changes for matcaps:
One is taking advantage of drawing with pbvh (which would only happen
with dyntopo previously) and drawing with partial redraw during
sculpting.
The second one is support for masks. To make this work in the special
case of multires, which uses flat shading, I use the only available flat
shaded builtins in OpenGL 2.0 which are color and secondary color.
Abusing colors in that way is also essential for flat shading to work if
we are to use pbvh draw in multires, since it is the color that is being
interpolated flatly, not the normal (which can only interpolated
smoothly). The pbvh drawing code for multires used last triangle
element's normal to compute the shading which would only produce smooth
results. This could change if we did the shading in the vertex shader
for flat shaded primitives, but this is more complex and makes it harder
to have one shader to rule the mole.
Also increased the brightness of the default diffuse color for
sculpting. This should be useful since artists like to tweak the
lighting settings and it will give them the full dynamic range of the
lights, but also it helps with correct brightness of sculpted matcaps.
Reviewers: brecht
Differential Revision: https://developer.blender.org/D435
When you press zoom 3 times you will now get 2:1, 4:1, 8:1, etc. Fixes T36916.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D409
Now they do, to make it harder to accidentally press them and lose work.
Reviewed By: brecht, carter2422
Differential Revision: https://developer.blender.org/D440
Issue is likely caused by thread-unsafe nature of IMB_freeImBuf
which might lead to race condition in some circumstances.
Now made it thread-safe and from Sebastian's tests seems crash is
gone now, so hopefully the root of the issue is finally nailed down.
* Fixing trailing whitespace in some files
* Fixing some indentation
* SCA_PythonController.owner is now documented
* SCA_PythonKeyboard members now use the member directive instead of function
Evaluation of time-warping modifiers ("Stepped" is one of them) didn't actually
end up distorting the time to look up what values other modifiers in the stack
generate. This meant that when a "stepped" fmodifier was on top of a "generator",
the stepped fmodifier looked like it didn't have any effect.
(This fix requires a bit of testing still, so should be left for 2.71)
- autodetect optimal default, which typically avoids HT threads
- can store setting in .blend per scene
- this does not touch general omp max threads, due i found other areas where the calculations are fitting for huge corecount
- Intel notes, some of the older generation processors with HyperThreading would not provide significant performance boost for FPU intensive applications. On those systems you might want to set OMP_NUM_THREADS = total number of cores (not total number of hardware theads).
Previously, amplitude was more of an "absolute" value in the sense that whatever value
you set it to became a sort of "maximum bounce" height. However, it turns out that this
approach isn't so nice when dealing with large gaps between the values of two keyframes,
as the elastic easing equations expect that "amplitude > |change|" (where change is the
difference in values from key1 to key2).
Now, the "amplitude" value we pass to the easing functions are "|change| + amplitude".
This is easier to control, as now, as soon as you start changing that value, there are
immediately visible effects.
Non-mesh objects default to no collisions. lordloki provided a
patch that sets the collision properties for a mesh converted from
a curve/meta/surf/text to the default collision properties for
a mesh.
This now uses decoupled ray marching, and removes the probalistic scattering.
What this means is that each AA sample will be slower but contain less noise,
hopefully giving less render time to reach the same noise levels.
For those following along, there's still a bunch of volume sampling improvements
to do: all-light sampling, multiple importance sampling, transmittance threshold,
better indirect light handling, multiple scatter approximation.
This basically records all volumes steps, which can then later be used multiple
time to take scattering samples, without having to step through the volume
again. From the paper:
"Importance Sampling Techniques for Path Tracing in Participating Media"
This works only on the CPU, due to usage of malloc/free.
Similar to surfaces, this will now always scatter rather than probabilistically
scattering or not depending on the transmittance.
This also makes calculation of branched path throughput non-probalistic, which
makes thing slower too. That's to be solved by decoupled ray marching later.
This removes a few optimizations to avoid exp() calls and division, they will be
added back later, at the moment it's more important to make the code easier to
understand and refactor.
Rather use random point in each step instead of giving the steps random sizes.
Gives a bit more accurate results with large step sizes, but also convenient
convention for later changes.
These can currently be accessed by adding an Attribute node and specifying one
of those three names. A Smoke/Fire node should be added at some point to make
this more convenient.
These values might change still before the release, in particular for flame the
meaning seems unclear, it's just values in the 0..1 range. This is useful for
color ramps, but it might be good if this was also available as temperature in
kelvin so it can be plugged into the blackbody node. But I couldn't figure out
from the smoke code if or how this corresponds to a physical unit.
Here's a (quite poor) example file for a fire + smoke setup:
http://www.pasteall.org/blend/27990
These are internally stored as a 3D image textures, but accessible like e.g.
UV coordinates though the attribute node and getattribute().
This is convenient for rendering e.g. smoke objects where data like density is
really a property of the mesh, and it avoids having to specify the smoke object
in a texture node, instead the material will work with any smoke domain.
Notes:
* The motion steps only affect deformation motion blur.
* The actual number of steps is 2^(steps - 1). This avoids having to sample at
many different times for object with more/fewer steps, now the times overlap.
* Deformation motion blur is enabled by default in existing files that have
motion blur enabled. If the object is not deforming, this will be detected at
export time, so raytracing performance will not be affected.
Part of the code is from the summer of code project by Gavin Howard.
This now supports multiple steps and subframe sampling of motion.
There is one difference for object and camera transform motion blur. It still
only supports two steps there, but the transforms are now sampled at subframe
times instead of the previous and next frame and then interpolated/extrapolated.
This will give different render results in some cases but it's more accurate.
Part of the code is from the summer of code project by Gavin Howard, but it has
been significantly rewritten and extended.
A question here might be: why?
The answer is that GPUs will convert this to triangles so we are wasting
memory during conversion of the element buffer to a triangle buffer
anyway.
The second reason is that some modern GPUs are slower rendering GL_QUADS
The third reason is that due to element caching, this should not in
theory be slower (cached elements do not get recalculated on the GPU).
- display handles with solid color.
- active handle is highlighted.
- 1/4 of the colorband draws solid.
- use old-style handles when zoomed out.
D289 by Charlie Jolly with own modifications.
Issue is that sampling functions did not pass a thread index to the
texture sampler so all threads were contesting for the same pool.
Paint cursors and sculpting that used openmp for threading suffered from
this. Now use omp_get_thread_num to pass the thread number.
Summary:
The title actually says it all, it's just possible to
have independent free handles for mask splines. Also
it's now possible to have aligned handles displayed
as independent handles.
Required changes in quite a few places, but they're
rather straightforward.
From user perspective there's one really visible change
which is removed Handle Type menu from the panel. With
asymmetric handles it's not clear which handle type to
display there. So now the only way to change handle type
is via V-key menu.
Rewrote normal evaluation function to make it deal
with new type of handles we support. Now it works in
the following way:
- Offset the original spline by maximal weight
- Calculate vector between corresponding U positions
on offset and original spline
- Normalize this vector.
Seems to be giving more adequate results and doesn't
tend to self-intersect as much as old behavior used to,
There're still some changes which needed to be done, but
which are planned for further patch:
- Support colors and handle size via themes.
- Make handles color-coded, just the same as done for
regular bezier splines in 3D viewport.
Additional changes to make roto workflow even better:
- Use circles to draw handles
- Support AA for handles
- Change click-create-drag to change curvature of the
spline instead of adjusting point position.
Reviewers: campbellbarton
CC: sebastian_k, hype, cronk
Differential Revision: http://developer.blender.org/D121
Root of the issues comes to the fact that it's possible to produce
a situation when library object data uses local object. This is
actually forbidden and not supported by .blend IO.
Made it so Make Local wouldn't produce such an unsupported states.
Reviewers: brecht
Differential Revision: https://developer.blender.org/D372
Bevel Factor Mapping allows to control the relation between bevel factors
(number between 0 and 1) and the rendered start and end point of a beveled
spline.
There are three options: "Resolution", "Segments", "Spline". "Resolution"
option maps bevel factors as it was done < 2.71, "Spline" and "Segments"
are new.
* "Resolution“: Map the bevel factor to the number of subdivisions of a
spline (U resolution).
* "Segments“: Map the bevel factor to the length of a segment and to the
number of subdivisions of a segment.
* "Spline": Map the bevel factor to the length of a spline.
Reviewers: yakca, sergey, campbellbarton
CC: sanne
Differential Revision: https://developer.blender.org/D294
Transform operators for nodes were not taking parent nodes (frames) into
account. Now use the nodeToView/nodeFromView functions to apply
transforms in local node space.
The ultimate goal is to only allow the rasterizer to handle OpenGL and bf_gpu
calls. This commit creates a RAS_ILightObject interface and a RAS_OpenGLLight
implementation.
These types of meshes do not use material caching, and thus only the first created
material would be saved, but subsequent ones were not. Those subsequent materials
were then not being freed. Now we make sure to track all of the materials.
Note: Meshes that cannot make use of material caching (no materials or using face textures)
can still use up a large amount of RAM since a material is created per face.
Fixes T39423: Blender crashes on iv41 encoded videos in thumbnail display mode
Libraries are still being committed to the svn, so compilation might be broken
for a while. Sorry for this, but can't really be avoided.
Also, some typos in scons are possible.
Issue here is complex (Of course, this is particles!)
First issue is that use_modifier_stack will use the num parameter of the
particles instead of num_dmcache, something the brush code did not
account for at all. Now correctly set DMCACHE_ISCHILD in that case.
Second issue is that make_derived_deform will return a mesh with less
indices than the particle system derived mesh. This would mean that
subsequent sampling of the particle derived mesh to initialize the
particles woould also produce garbage. This was being done for
optimization but in that case it broke the system.
Reviewers: lukastoenne
Differential Revision: https://developer.blender.org/D429
do not use different stroke property names for different paint systems.
This was done due to different stroke sets being supported for each
system, but this lead to trouble if we changed the names (due to
different stroke sets being supported) and users created custom keymaps
with the old property name saved.
The first part of this fix addresses master. A similar commit will be
done to soc-2013-paint.
which tiles are selected for input there was always a constant for correcting
the accuracy.
It seems that the constant was not enough and has been adjusted. (2 => 3).
BKE_sequencer_offset_animdata() was simply assuming bezt member of fcurve was always valid, while it might be NULL (e.g. when fcurve is using FPoints instead, like when generated from sound file).
added in rB74518b28267e9b18199212fbaa3c689fa018d20c.
No special bind/unbind needed for standalone viewer, so can just use a
static stub in the display callback.
Issue was caused by the wrong usage of OCIO GLSL binding API. To make it
work properly on pre-GLSL-1.3 drivers shader is to be enabled after the
texture is binded to the opengl context. Otherwise it wouldn't know the
proper texture size.
This is actually a regression in 2.70 and to be ported to 'a'.
Compute barycentric coordinates (u, v) for a point with respect to a
triangle.
This is needed for Cycles baking but we decided to push this
independently of the upcoming main baking changes.
Code adapted from Christer Ericson's Real-Time Collision Detection.
Cleanup, refactoring and review from Campbell Barton.
Following problems:
- painting onto a cube asserts immediately (vpaint/wpaint)
- crashes with valgrind (bad memory use)
- the original intended behavior of vertex projection paint no longer works as I intended.
(its constantly refreshing geometry even when no modifiers are applied).
Best get code review for these kinds of changes.
Crashes due to bad dm face calculation in particle RNA/cycles hair.
This stuff is a total mess, but this kind of check is used in other
places and prevents crashing at least.
Issue was caused by inverting a degenerate matrix when
evaluating drivers.
Solved by using tweaked inverse code (same as used in Cycles).
Should have no affect on cases when matrix is not degenerate.
Using unordered_map and unordered_set C++ container types currently
requires careful testing or usage of boost, due to the various confusing
C++ version differences in include paths and namespaces.
Libmv defines tests for these cases in cmake and scons, such that ceres
can use any available implementation, or fall back too std::map/std::set
if none can be found.
This patch generalizes this buildfile code by providing a Blender macro.
* cmake: defines both the variables used by libmv at them moment as well
as 2 variables UNORDERED_MAP_INCLUDE_PREFIX and UNORDERED_MAP_NAMESPACE,
which can later be used in other C++ parts for convenience.
* scons: adds a tool script returning the include prefix and namespace.
Libmv checks these to define the appropriate definitions for ceres.
Differential Revision: https://developer.blender.org/D425
Issue was caused by the cache limitor which was removing 4k textures from the
memory when accessing other images.
This is pretty much awful situation and solved by making it so only image sequences
and movies ace cache-guarded.
Could be optimized further so images used by viewport are not being freed, but
that's much more tricky to do..
This is a nice candidature for 'a'.
If bge.logic.addScene() could not find the scene to add it would add the
first scene again, which is just silly. Now, if no scene is found, a warning
is printed and nothing is added.
* Factor values get a minimum circle too, which denotes the maximum
value. This makes it easy to set the maximum value while previously
maximum was at center, much more difficult to set exactly.
* Added text indication of value at center of the widget.
There are a couple of bugs that come together here:
* Particle hacks: extra modifier stack evaluation just for particles in
rna_Object_create_duplilist. This is where the primary issue stems from,
the "for_render" setting replaced the G.is_rendering flag in threaded
depsgraph. This causes particles to recalculate the entire modifier
stack with _render_ settings instead of viewport settings now. Fixed by
taking the 'preview' parameter in Cycles into account.
* Buggy skin modifier: The skin modifier generates a different amount of
vertices and faces **on every execution**. This must be looked at
separately, but it could be another reason why cycles constantly
restarted the sync process.
* Particles get re-distributed randomly every time (changing seed). This
could be caused just by the broken skin modifier, but might still be an
issue when simply rendering with cycles, since the psys will be
evaluated for render settings, if just temporarily.
Outliner rename callback is supposed to activate affected element before actually renaming,
but for bones this was not working because the function used to activate the object explicitely
ignored ID_OB case! Added a bool flag to allow handing this case without (possibly) breaking
the other usecases.
Issue is a regression since threaded objetc update and caused
by the fact that some objects might share the same proxy object.
It's all fine but object_handle_update() will call update for
a proxy object which screws up threaded update.
The thing is, proxy object is marked as depending on a scene
object and such a call makes it so the children objetc is
being updated.
This is really bad and depsgraph is to take all responsibility
on updating the proxy objects.
So for now used a simple solution (which is safe to backport
to 'a') which is skipping proxy update if the scene update is
threaded and based on the DAG traversal.
There are some still areas which calls object update directly
and for that cases proxy object is still being updated from
object_handle_update().
Ways how it was resetting its values (backspace) was far from satisfaying. Now, e.g. when scaling, it will reset at 1 (or whatever mouse-value it was before entering numinput), instead of some ugly 0.0 value.
Implementation details:
* Values passed to applyNumInput() are stored as default ones (val_org), if it is not EDITED.
* applyNumInput() returns a boolean saying whether it actually set values or not.
* When backspace hits its ultimate step (where it clears all EDITED flags and reset all default values),
it sets a temp FAKE_EDITED flag that will be used to apply one last time values of numinput
(so that default values actually get applied!).
There are important things to note here for code using numinput:
* Values passed to applyNumInput() should be valid and are stored as default ones (val_org), if it is not EDITED.
* bool returned by applyNumInput should be used to decide whether to apply numinput-specific post-process to data.
* *Once applyNumInput has been called*, hasNumInput returns a valid value to decide whether to use numinput as drawstr source or not.
Those two steps have to be separated (so do not use a common call to hasNumInput() to do both in the same time!).
Located on topology panel.
To use just click on button and click on mesh.
Operator will just use the dimensions of the triangles below to set the
constant detail setting.
Also changed pair of scale/detail size with nice separate float
percentage value.
Gives ~11% speedup for hair.blend, ~10% for koro_final.blend
Also extract few common subexpressions in hair calculation.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D318
Nothing spectacular here, fill tools are easy. Just take the dyntopo
code and repeat until nothing more to do.
The tool can be located in the dyntopo panel when the dyntopo constant
detail is on.
Also added scale factor for constant detail. This may change when detail
sampling gets in, I am not very happy with having two numbers here,
still it will give some more control for now.
Inspired by T39315, this commit adds a few more driver "linting" messages used for
providing users with tips on how to use drivers better. This time, we specifically
address 2 cases:
1) Drivers being abused for procedural animation, due to the misconception that
procedurally generating F-Curves using F-Modifiers means that drivers are needed
to wire such procedural motion-sources to properties.
2) Setting up Average/Sum/Min/Max driver types without any input variables - you can't
expect anything to happen (unless of course, your intention was to lock the property
to 0.0)
Issue here is that coordinates used for projection would not fit the
closest candidate. So it was possible to paint against the mirrored part
of the mesh, which would get clipped on reprojection.
Fix avoids reprojection by storing squared distance, and screen space
coordinates, which we will need again later anyway. Also we now always
paint against the closest vertex always. This is ensured because squared
distances for uninitialized vertices will be MAXFLOAT, thus failing the
strength test.
- i empirically found that using HT just gives overhead, even possibly slowing things down
- i propose other OS test this too again
- switching threaded on/off did not restored single threaded case ( credits to Sergey )
This commit introduces support for a number of new interpolation types
which are useful for motion-graphics work. These define a number of
"easing equations" (basically, equations which define some preset
ways that one keyframe transitions to another) which reduce the amount
of manual work (inserting and tweaking keyframes) to achieve certain
common effects. For example, snappy movements, and fake-physics such
as bouncing/springing effects.
The additional interpolation types introduced in this commit can be found
in many packages and toolkits (notably Qt and all modern web browsers).
For more info and a few live demos, see [1] and [2].
Credits:
* Dan Eicher (dna) - Original patch
* Thomas Beck (plasmasolutions) - Porting/updating patch to 2.70 codebase
* Joshua Leung (aligorith) - Code review and a few polishing tweaks
Additional Resources:
[1] http://easings.net
[2] http://www.robertpenner.com/easing/
as well.
These were already doing the same thing, just not as nice. Only
difference is the do_action argument (false for BKE_libblock_copy_nolib)
but this is of no consequence because the function is only called for
trees nested inside material, scene, etc., which never have own actions.
depsgraph updates.
Material datablocks were localized by first making a regular datablock
copy, which always gets inserted into the bmain list, and then removing
it again from bmain.
Problem is that this localization happens in preview threads, which can
run while the depsgraph is also updating GPU materials. In case the
copying of materials takes any amount of time, this can cause the
depsgraph call to material_changed to use an invalid, localized material
and access invalid GPUMaterial lists which have already been freed for
the actual material.
Solution is to not add localized datablocks to the bmain lists in the
first place. bmain should be totally immutable during preview or render
threads.
Trying to use mat game settings in 3DView makes sense - but only when we are using "Game Engine" as 'renderer'!
Note GLSL code completely ignores mat game settings currently, could be added too...
Yet another attempt at fixing the problems here. This time, I've added a new
function/version of the binary search utility so that we can pass in custom
thresholds (Note: This ability is only used for evaluation currently, with
everything else using a wrapper which still uses the old default threshold),
making it ok to start trusting the "exact" parameter.
This is mainly a maintaince commit which syncs changes
between Blender and Libmv upstream also bringing new
GLog version.
This GLog version is presumably have better support of
MinGW from "the box".
This commit is also aimed to make further 3d part libs
update easier.
the inline definitions are included, which gives lots of warnings
"attribute declaration must precede definition".
See also rB83617429cf28f2a19e991a0f71d892fc159a4419
Basically issue was caused by the fact that strip for proxy has been
post-processed but proxy files were considered pre-processed. This lead
to situation of postprocessing being applied twice.
GPC_Canvas::GetWindowArea() and KX_BlenderCanvas::GetWindowArea() are tracking two
different things. The 2d filter system doesn't actually need the window area and
can get by just using the viewport dimensions, which it now does. The GetWindowArea()
mess should probably be cleaned up.
This commit attempts to fix some of the FCurve evaluation regressions arising from
an earlier commit to speed up the process using binary search. Further tweaks may still
be needed though to get this to an acceptable level of reliability (namely, tuning the
threshold defining which keyframes get considered "close together"). Since we're still
in an early stage of the 2.71 dev cycle, for now it's still worth trying to get this
working instead of simply reverting this (which can still be done later if it proves too
problematic).
Specific fixes:
* The previous code was somewhat dangerous in that it allowed out-of-bounds accessing
of memory when a == 0. It turns out this was more common than originally anticipated
(as the assert I added here ended up failing in the "action_bug.blend" file in the report)
* Tweaked the code used to test for closely-spaced points so that the "Clive.blend" example
for driver curves won't fail. The approach used here has the downside though that
since "exact" uses a might coarser threshold for equality, there may be some precision
loss issues causing backwards compat issues (namely with closely spaced keyframes, or
for certain NLA strips).
For now, I've left in some debug prints that can be enabled by running Blender in debug
mode (i.e. "blender -d"), which can provide some useful tuning info should we need to
look more into our approach here.
When path to resolve "finishes" on a collection prop, do not erase the returned prop!
This caused py's path_resolve to return same PointerRNA as the one passed as parameter, leading to
inifinte recursion in Operator's accessor func (__getattribute__)...
When leaving cam view, if autopersp is ON, and previous view was an axis one, we have to force new view in perspective mode - else, we reuse previous stored mode (persp or ortho).
Fluid sims have a very nasty feature for interaction, in which a psys
can directly update the bvhtree for //another object's psys//. This
breaks with threaded depsgraph evaluation and is generally a no-go.
To avoid crashes for now, use a global mutex to avoid concurrent writes
to an object psys' bvhtree.
The issue was caused by assumption object matricies can be modified
by DupliObject->obmat outside of the duplilist iteration cycle.
This isn't really true. The thing is, old code used to modify object
matrix based on the duplicated matrix to make it so modifier evaluation
kind of works (it'll use proper matrix in most cases, but think there
might be some cases when this assumption also breaks).
Even when matrix is copied from dupliobject to object in advance
(to mimic old behavior or dupli system) you're still to use matrix
from dupliobject inside a cycle body. This is because the same object
might be modified multiple times by different dupliobject. For example,
in case of duplifaces loads of faces will modify object's matrix and
before this change matrix from the last face would have been used.
The fix is simple one-liner which makes it so render instance will
use proper matrix from dupliobject.
All the rest bits here (pre-loading matrices and restoring them in
the reverse order) totally matches old behavior of duplilist.
This is a regression and to be ported to 2.70 tag!
Make HSL wheel useful in compositing.
In HSL, 0.5 lightness corresponds to most vibrant colors so make that
default for locked wheels (previous value of 1.0 just made it white).
Also, unlock the wheel for both very bright and very dark colors.
and value/lightness slider stops midway.
Issue here is manyfold:
Color wheel does not support properties with different soft min/max
values than 1.0 (which after experimenting a little I left as is), and
also color management is completely destroying the mapping between the
value slider and the RNA property value range. To solve this I have
disabled color management by setting the property to gamma corrected
(only in RNA, Sequence editor coders please check!), otherwise it will
just become a big mess of tracking where color comes from and what kind
of color transforms it needs in different color pickers (if property has
non normalized range etc). HSL is not really meant to represent colors
outside a normalized space so I have disabled setting lightness above
1.0 in this model.
This will work, however it is hacking a color picker to do something
other than what it is supposed to do: pick a color from the screen
accurately. Which means normalized values always. The non normalized
colors picked for lift/gain/gamma through the pickers do not correspond
to any accurate colors; they are rather a user friendly way to 'sort of'
choose a color and a gamma with an indication of maximum value.
I think that lift/gamma/gain nodes need a dedicated widget for this
(besides it is quite clear that some options are written for that use
case) -or- a separate gamma multiplier for the picked color (which
should itself be in a normalized space)
Was using wire or black in many places, this color is used for cursor,
camera guides, transform helper lines. So its possible to have a dark
background with light overlay color.
Patch D331 by Brita, with some edits.
Issue was in BLI's rotation_between_vecs_to_quat(), which did not handled correctly cases where both vectors are colinear.
Patch by Campbell Barton and me.
Issue originaly tracked down by Yan Shi, many thanks!
This is a regression caused by rB67134a7bf689279785e2e40b29cd24243813998b
The UV coordinates read from the UV input must be scaled by the Image
input size instead of the UV input size.
Also now this node uses the UV input resolution instead of the Image
resolution, since this is what determines the available resolution. The
image is EWA-sampled anyway, it's resolution does not have a direct
impact.
This provides a speedup to evaluating long F-Curves in fcurve_eval_keyframes()
by using the pre-existing binarysearch_bezt_index() function (used for keyframe
insertion) to find the relevant BezTriple on the FCurve at the current evaltime.
The current code loops over all BezTriples (sometimes not even breaking from the
loop after cvalue has been evaluated).
Reviewer Notes:
- Unlike in the original patch, we use the old/existing logic instead of
checking that (exact == true). See comments in code and also on the tracker
entry for this patch for more details.
Patch By: Josh Wedlake
To reduce user confusion, we were already presenting users with an error message
when they tried to edit constraints for bones from the Object Constraints tab.
This commit just makes things more convenient by adding a button which takes
users to the right tab when clicked.
This is a failure of viewport direct displist creation
caused by existing curve_cache pointer which empty content.
Made it so if the curve isn't evaluated it's curve_cache is NULL.
This is just-another-regression to be ported to the release.
Basic idea is to check whether an element is visible or not, and ignore those that are hidden, during move up/down.
Reviewers: aligorith
Reviewed By: aligorith
Differential Revision: https://developer.blender.org/D376
Fluid particles use the particle system's bvhtree structure, which is a
runtime BVH tree. This was not reset properly on copying objects/psys,
which lead to concurrent access in threaded depsgraph updates and memory
corruption.
This is mostly useful for precision in lift/gamma/gain nodes. Display of
those values is done with linear interpolation while saturation is
interpolated cubically. This will lead to color inconsistency between
selected and displayed color on color pickers for instance, which should
be avoided on those use cases.
This was supposed to fix inaccuracies between S-H wheel but it makes the
colors selected through the SV-H color cube wrong (see http://
%28SV-H%29-Inaccurate). Instead of deactivating color management for SV
and H widgets just reenable correction for the Hue wheel. This is wrong
because HSV always refers to display space and we -have- to account for
that when using linear color RNA properties.
history operator (Ctrl + Alt + Z).
This will only show paint operations now while in an image paint mode.
The caveat is that user can delete previous paint operations too (even
on images not on the canvas currently) so it needs some care. This is
consistent with regular undo behaviour though.
Sculpting also suffers from lack of Undo history support, this will be
added in a separate commit.
Core issue exists since ages - the thing you get from bpy.types.YOUR_OT_operator is fuzzy, and may change,
due to the fact that both Operator and OperatorProperties share the same name...
Would be cool to get rid of this issue one day, but for now it's safer to use rna acessor...
This commit is to be backported to 2.70.
We can't simply dupalloc cache_system (LaplacianSystem), it has quite a few allocated data we'd need to dupalloc as well,
not to mention (nl) context... Much safer to just set it to NULL in new copy imho!
This commit is to be backported to 2.70 release!
This re-applied patch from 25cbd13 which was lost at some point
since missing patchset in series.
This revision is to be back-ported to the final release.
Own regression since partial buffer update refactor,
display buffer for exr tiles should happen in image
update callback.
This revision is to be merged to the final release.
The previous fix 162d6c73e3 has been cancelled because it caused a
problem that mesh deform cages and so on mess up the preview. To avoid
the problem, viewport doesn't show mesh object if its draw type is
bounding box or wireframe.
Reviewers: brecht
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D375
This is a standard Hue - Saturation - Lightness model
(see for instance entry on wikipedia here: https://en.wikipedia.org/wiki/HSL_and_HSV)
Note though the difference between HSV and HSL saturation, which are not the same.
The advantage of having this color selection scheme is that artists can select
shades and tints of a color easily by using the lightness slider. Also colors
are arranged on (approximated) perceived lightness on the color wheel.
Beware, Old files opened with this preference saved will crash!
Reviewers: sergey, brecht, campbellbarton
Differential Revision: https://developer.blender.org/D385
I overlooked in cmake my fix same time changed the optimization level to 0,
so not use optimization is the real fix ( postponed for after 2.70 ).
I appears we should investigate scons anyway: compileflags does not apply to c and c++ same time as expected.
Enum icon-only buttons were getting their strings set,
then truncated with blenders string shortening methods, then not drawn
because there was no room (since buttons are icon width).
Modify UI code so icon-only buttons don't get names and passing "" to a
button won't have its text set later on.
This was caused by static variables used in plNearestPoints().
For now solved by making the solvers allocated in the stack,
seems no noticeable affect on the simulation speed so far.
This is a regression introduced in fbc7ab30f and caused by the wrong object
being passed to the function.
Not sure passing editnurb as an argument is a good thing to do..
The copy-to-selected operator for RNA buttons uses paths for copying
object pointer properties. Copying other ID data blocks is deliberately
disabled:
https://developer.blender.org/diffusion/B/browse/master/source/blender/editors/interface/interface_ops.c$274
However, the RNA_path_resolve_full function is not properly working for
retrieving pointer properties: it always will dereference pointer
properties in anticipation of further path elements.
In fact the return value of RNA_path_resolve_full has a conflicting
double meaning. It returns `false` when
* the RNA path is invalid
* any of the pointer properties is NULL
This means that it is not capable of returning pointer properties at all.
To make this possible, there is now an internal function for path
parsing, which returns false //only// if the the path is invalid.
On top of this there are 4 wrapper functions for retrieving either
actual property values (RNA_path_resolve, RNA_path_resolve_full) and for
retrieving pointer+property pairs (RNA_path_resolve_property,
RNA_path_resolve_property_full). The latter 2 variants will **not**
dereference pointer properties at the end of the path, so callers can
actually get the property itself. The `***_full` variants include an
array index return value.
Differential Revision: https://developer.blender.org/D396
This was suggested by Christopher Barrett (terrachild). Corner pin is a common feature in compositing.
The corners for the plane warping can be defined by using vector node inputs to allow using perspective plane transformations without having to go via the MovieClip editor tracking data.
Uses the same math as the PlaneTrack node, but without the link to MovieClip and Object.
{F78199}
The code for PlaneTrack operations has been restructured a bit to share it with the CornerPin node.
* PlaneDistortCommonOperation.h/.cpp: Shared generic code for warping images based on 4 plane corners and a perspective matrix generated from these. Contains operation base classes for both the WarpImage and Mask operations.
* PlaneTrackOperation.h/.cpp: Current plane track node operations, based on the common code above. These add pointers to MovieClip and Object which define the track data from wich to read the corners.
* PlaneCornerPinOperation.h/.cpp: New corner pin variant, using explicit input sockets for the plane corners.
One downside of the current compositor design is that there is no concept of invariables (constants) that don't vary over the image space. This has already been an issue for Blur nodes (size input is usually constant except when "variable size" is enabled) and a few others. For the corner pin node it is necessary that the corner input sockets are also invariant. They have to be evaluated for each tile now, otherwise the data is not available. This in turn makes it necessary to make the operation "complex" and request full input buffers, which adds unnecessary overhead.
option after reentering a paint mode.
Solution by Bastien with modifications, thanks!
Show Brush flag need not be reenabled always, but make sure it is at
least enabled once on paint initialization.
* Moved collection key parsing and array index parsing into their own
functions, to make the main path loop easier to follow.
* Unified boolean return values.
This adds a new option "Sample All Lights" to the Sampling panel in Cycles (Branched Path). When enabled, Cycles will sample all the lights in the scene for the indirect samples, instead of randomly picking one. This is already happening for direct samples, now you can optionally enable it for indirect.
Example file and renders:
Blend file: http://www.pasteall.org/blend/27411
Random: http://www.pasteall.org/pic/show.php?id=68033
All: http://www.pasteall.org/pic/show.php?id=68034
Sampling all lights is a bit slower, but there is less variance, so it should help in situations with many lights.
Patch by myself with some tweaks by Brecht.
Differential Revision: https://developer.blender.org/D391
In many BI shader setups, the material which owns the nodetree is often itself
included as a node in that nodetree (i.e. to provide the base colour for that
mesh). This would often result in the material (and its subtree) getting included
in the dopesheet results twice.
This can freeze Blender if you accidentally press ctrl+m on a large
script, conversion from text to 3d objects is fairly special case,
restricting access to the menu only.
When the keyframes at either end of the source curve don't lie on exact frame boundaries,
this casued problems with the Cycles F-Modifier, as part of the cycle would get chopped
off.
This was caused by float -> integer truncation that was occurring, since one variable
was of the wrong type. The problem here wasn't discovered until now (thanks to gcc's
invalid-type warnings on printf's) as in standard usage, we can safely assume that all
keyframes are strictly on frame boundaries.
In order to combat the problem of users frequently trying to use inlined
bpy.data/bpy.context paths for data access in their driver expressions
and then finding/complaining that these don't update correctly, the UI
now flags these as the error conditions that they are (with suggestions
on how to fix this).
Also tweaked the "Add Variable" button to have more descriptive text about
what exactly variables are and why to use these, along with some other
visual tweaks (icons!).
All textures are sampled bi-linear currently with the exception of OSL there texture sampling is fixed and set to smart bi-cubic.
This patch adds user control to this setting.
Added:
- bits to DNA / RNA in the form of an enum for supporting multiple interpolations types
- changes to the image texture node drawing code ( add enum)
- to ImageManager (this needs to know to allocate second texture when interpolation type is different)
- to node compiler (pass on interpolation type)
- to device tex_alloc this also needs to get the concept of multiple interpolation types
- implementation for doing non interpolated lookup for cuda and cpu
- implementation where we pass this along to osl ( this makes OSL also do linear untill I add smartcubic to the interface / DNA/ RNA)
Reviewers: brecht, dingto
Reviewed By: brecht
CC: dingto, venomgfx
Differential Revision: https://developer.blender.org/D317
dyntopo
Layer brush would not invalidate the layer_disp arrays in dyntopo mode,
checking only for the existence of the array. This means that if a tool
resized the node due to topology changes, the layer brush code could
index (and write!) out of bounds in the array. Solution is to invalidate
the layer data prior to each stroke in dyntopo.
- Added update callback to perform on-update validation when changing the
extrapolation mode on F-Curves
- There was a patch in the tracker for adding an "update()" method to F-Curves
which does a similar thing when manually called by scripts. Since we've added
a function for this in RNA anyways, we might as well add this too while we're
at it. (NOTE: upon closer inspection, the original patch by Tom Edwards had a
number of issues, so I ended up reimplementing here)
in threaded depgraph updates and effector list construction.
Gathering effectors during depgraph updates will call the
psys_check_enabled function. This in turn contained a DNA alloc call
for the psys->frand RNG arrays, which is really bad because data must be
immutable during these effector constructions.
To avoid such allocs the frand array is now global for all particle
systems. To avoid correlation of pseudo-random numbers the psys->seed
value is complemented with random offset and multiplier for the actual
float array. This is not ideal, but work sufficiently well (given that
random numbers were already really limited and show repetition quite
easily for particle counts > PSYS_FRAND_COUNT).
There were 3 bugs with both data types
- using freed memory while sorting.
- sorting failed in some situations.
- scaling allowed multiple items to be on the same frame.
Replace this with a simple sort + de-duplicate, taking selection into account.
Now Alt+RMB,RMB will select all connected boundaries.
There are times when you just want to select an entire boundary loop
ignoring face topology, previously there was no way to do this.
This switches api usage for cuda towards using more of the Async calls.
Updating only once every second is sufficiently cheap that I don't think it is worth doing it less often.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D262
Dyntopo detail in object space. This allows to set the detail in
percentage of blender units and sculpt in this detail constantly,
regardless of the distance to the mesh.
This commit just enables the functionality, which is really trivial.
There will be some more commits like detail flood fill and
detail sampling in the future.
The problem here is that the redraw rectangle can be calculated many
times before actual redraw (for instance due to repetitive calls to the
paint step function for space strokes). Those redraw rectangles need to
be accumulated and finally flushed when the draw call is done.
Also, it is this accumulated rectangle that must account for the
previously drawn rectangle to avoid artifacts due to drawing.
Now all looks like it works correctly :)
after the stroke so that it can be unioned with current rect correctly
during the draw update. For some reason dyntopo clipping misbehaved
before that.
For very thin faces, knife sometimes missed cutting a few.
Problem was that a test for edges being totally inside a face
was being applied in inappropriate circumstances. An assumed
invariant about the knife hit structure is not true when
actually in the middle of making cuts.
Also make a 'big epsilon' a little smaller, though don't think
that was the problem in this bug.
* Do union with previous partial redraw before overwriting with self and
store unmodified new rect to avoid rect getting bigger.
* Make rect one pixel wider instead of smaller so edges do not get left
outside. Overall this looks quite better.
Basically proxy colorspace didn't work well enough.
It is still a bit weird and mainly:
- Proxies for image sequences are built in the image color space.
- Proxies for movies are built in the movie color space.
This could be unified but would need some work in proxy build
to make it not just pipe frames from one FFmpeg context to
another but also apply OCIO on it.
- Addons and locales now points to v2.70-rc tag
- Also updated hashes for addons_contrib and scons
to make them up-to-date while i'm on updating this
hashes.
Testing for custom data availability with getVertData and index 0 fails
on assert for empty meshes (the index must be in valid range). Better
use getVertDataArray which has no requirement for specific index.
The problem is that comparisons involving the constants Nature.POINT (for vertices) and
Nature.NO_FEATURE (for edges) were evaluated in a wrong way. It is recalled that the
Nature class is a subclass of Python's built-in int type, and that these two constants are zero
when evaluated as numbers. The issue was caused by the implementation of the constants
in an incompatible way for comparison with Python int (and boolean) values. Specifically,
the zero of Python int is represented by an empty array of digits, whereas the zero-valued
Nature constants were represented by an array of size 1. Python int comparison operators
first check the lengths of the arrays of two operands, and then start comparing the digits
only when the array length is the same. For this reason, the two Nature constants were
not properly compared with int values (and thus with boolean values). It is noted that the
zero-valued Nature constants may result from bitwise operations on other Nature constants
(e.g., Nature.SILHOUETTE & Nature.BORDER), so this issue must have affected many
existing style modules.
The problem was reported by Folkert de Vries (flokkievids) through personal communications.
Thanks a lot!
Last change to bevel had a check for what was supposed
to be an "on edge" new vertex being off the edge.
The test tolerance was too small. This fixes that.
There were loads of issues in the code still which are mow likely fixed:
- Hole resolver hook had memory leak -- it didn't free face with holes
when triangulating it.
- Original edge mapping didn't work correct. old code related on the fact
that loop order is not changing when constructing the MeshSet class, but
in fact it does change.
Currently used edge map for this because it was easiest way to do it now,
but after the release we're to change it. Main reason is that face mapping
is not correct as well (and it was never correct actually). So we'll need
to construct Mesh structures by our own to be sure we're using correct
original index mapping.
- Carve might produce faces with ears, which is forbidden in Blender.
it wasn't an issue in old integration because triangulation will remove
the ears. So for now simply added ears removing back as a hook.
But actual reason of the ears is to be investigated really.
This hook will only work for NGons, quads are assumed not be able to
have ears. So this additional hook shouldn't slow down things much.
- Carve's hole resolver produces duplicated faces in some cases. Still not
sure what is the reason of this. Code here is not so much straightforward,
this is to be investigated later.
For now solved the issue as own hole resolver which checks for duplicated
faces after the hole resolving.
The additional checks here will only run if the mesh actually have hole
and wouldn't introduce slowdown for faces which doesn't have holes.
- Made it so if edge user triangulation gets a split (for example, in cases
when this edge intersects with the second operand) it wouldn't be dissolved.
This prevents cases of crappy topology after dissolving in several cases.
- Edge dissolver didn't check for whether edge is a non-manifold. We couldn't
really dissolve open manifold edges.
The bad thing about this is that mesh triangulation might produce non-manifold
edges and they wouldn't be dissolved. Not worst case in the world, but would
be nice to have it solved somehow.
- Exporting mesh form Carve to Blender might have produced duplicated edges
in cases when several non-manifold faces shared the edge. This is also fixed
now.
- Mesh triangulation might have produced duplicated faces, which is really bad.
Fixed by keeping a track on which faces we've created and skipping adding new
triangle if we already have one.
This all might introduce some slowdown, but we're too close to the release now,
so would rather have it slower but robust. After the release we might look into
ways to speed things up.
This made loops in linked lists of uvedges from duplicate members
causing a hang. Now check all previous links before adding a new one. It
is not much slower in practice because non-manifolds are not that
frequent.
to wrong obmats of nested dupli objects.
Restoring obmats after BI nested dupli object rendering has to happen
in reverse order, so higher level omats are applied last.
If you have two overlapping NLA strips it automatically blends between them.
However it was also blending between the first strip and the rest pose
(e.g. 0,0,0 for locations and 1,1,1 for scale).
This is ok if the blend mode is Add or Multiply since then you are adding onto
the rest pose, but for Replace you want to be able to mix between two poses
and not take the rest pose into account at all.
What this does is that to not do any blending for the first strip if it has
blend mode Replace. If someone wants the old behavior (from bugfix commit
27d792fa9c) they can still make an action with the rest pose and use that
as the first strip.
Grease pencil straight line can still be enabled by first pressing D
then Ctrl and looks like there is no final consensus on key shortcut.
Better leave this as before and solve after release.
Implements a more flexible internal connect function for standard nodes
(compositor, shader, texture). Allow feasible datatype connections by
priority.
The priorities for common datatypes in compositor, shader and texture
nodes are encoded in a simple function. Certain impossible connections
(e.g. color -> cycles shader) are excluded by giving them -1 priority.
Priority overrides link status: If a higher priority input can be found,
this will be used regardless of link status. Link status only comes into
play for inputs with same priority.
Reviewers: brecht
CC: sebastian_k
Differential Revision: https://developer.blender.org/D356
Issue of this bug is that most part of fftw is not thread safe, only compute-intensive fftw_execute & co are.
Since smoke was affected by this issue as well, a global fftw mutex was added to BLI_threads.
Audaspace also uses fftw in one of its readers (AUD_BandPassReader.cpp),
but this is not an issue currently since this code is disabled in CMake/scons files.
There was another threading issue with smoke, we need to copy dm used by emit_from_derivedmesh(),
as it is modified by this func.
Reviewers: sergey, brecht
Reviewed By: brecht
CC: brecht
Differential Revision: https://developer.blender.org/D374
There needed to be a check that when a newly created point is
supposed to be on an edge, that it stays within the bounds
of either end of the edge.
This fixes the hole-in-cube example in the bug, but not
the boolean modifier one, which still needs more work.
Actually, was broken for any custom modifier name, since it was explicitly using 'Cloth' one. Changed to mimic other cloth pressets (wonder why this one was different!).
Just remove that rotation special case for now, at least fixes the glitch along Z axis when rotating around Y axis in report.
Anyway, there is no reason for such special handling, we do not have real rotation in editbone...
This case was handled specially in writeffmpeg.c and seems it makes
audio export happy in all cases now.
TODO: libav-10 doesn't work with AC3 codec yet because this bloody
library ONLY supports FLTP format and FFmpeg ONLY supports FLT.
This is not fun guy, it really isn't! Where's your conscience??
CC: nexyon
Issue was caused by the way how audio output works from audaspace.
Now made it much closer to what's happening in ffmpeg.c and writeffmpeg.c.
Also fixed issues with incompatible combinations of codecs and formats
in mixdown settings.
Commit 162d6c73e3 changed the behavior of rendered viewport to use
viewport visibility, but that can cause some problems. For example,
mesh deform cage is drawn as a solid/textured mesh (not a wireframe
mesh) and its unnecessary surfaces and shadows mess up the preview.
The script ##cmake_linux_install.sh## is currently invoking ##make## in single-threaded mode; this patch changes it to take advantage of all available CPU threads.
Reviewers: mont29
Reviewed By: mont29
Differential Revision: https://developer.blender.org/D358
MSVC 2008 ignores alignement attribute when assigning from unaligned
float4 vector, returned from other function. Now Cycles uses unaligned
loads instead of casts for win32 in x86 mode.
This is needed to minimize their reprojection error over the footage.
Without this extra step positions of such tracks were calculated by
algebraic intersection code only, which doesn't give best precision.
This is a resubmission of the original patch from D255. Sorry, I didn’t understand that subsequent patches added to a diff are considered to //override// previous ones, rather than add to them.
Basically the comment for commit rB554eca1c288e has been applied to the wrong patch.
Reviewers: mont29
Reviewed By: mont29
Differential Revision: https://developer.blender.org/D359
Avoid zero-sized problem when doing euclidean intersection
Zero-sized problem might occur when intersecting track with
constant zero weight. For such tracks we'll just use result
of algebraic intersection.
TODO: We probably need to have a separate BA step to adjust
positions of tracks with constant zero weight.
There is a key conflict between grease pencil shortcuts and sculpt
shortcuts (D and ctrl D were taken by draw brush and dyntopo toggle,
respectively). Based on feedback, change dyntopo toggle to ctrl-T and
draw brush to X.
Also add missing property update for dyntopo detail size.
Note this commit only makes search menus wider and adds shortening to their strings.
Adding tooltips here is not trivial, would need much bigger changes...
Reviewers: brecht
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D341
This commit hopefully fixes all glitches we had when bone was Z-aligned. Note that when you init a transform
with a Z-aligned bone and change it to be non-Z-aligned, you will still get some brutal roll change,
there is not much things we can do here afaik...
renders were broken.
This was caused by rB1a79abdad2443ff9f12e7efd95ee78a264a9d60a which
makes a copy of the render layer list for thread safety. The single
layer passed to this function is still in the original list though, so
to get the correct index it has to be looked up there. Otherwise no
active index is set and all layers are rendered every time.
It would include/exclude shadow depending on the pass being disabled/enabled,
but that should have no influence on the combined render result. Now it always
includes shadow.
The render operator invoke checks render layers, which can force the
render layer to be activated. This requires a notifier, which has to be
done in the operator itself (can't do this inside pipeline code).
It doesn't make sense to use zero-weighted tracks as a correspondences
in keyframe selection.
Such tracks are not guaranteed to be tracked accurately because their
purpose is to add reference points in 3D space without affecting the
solution.
Problem is that the particle systems in the cycles database are not
stored in a well-defined order. This means the particle_id for dupli
objects can not simply be assigned using a global running index during
sync.
Now the particle index is assigned locally for each particle system.
When transferring particle data to the device as a single texture, the
particle indices are offset based on the final order of particle systems
in the database.
Reviewers: brecht
Reviewed By: brecht
CC: Andreas80
Differential Revision: https://developer.blender.org/D352
This gives much worse results on mango footage (see 04_2e)
so disabling for now for until proper prediction model is landed.
The thing is, currently blender sends input coordinates as the guess to
region tracker and in case of fast motion such an early out ruins the track.
Issue was caused by cycles setting scene frame which will update scene for
all the layers (not just visible ones) which confuses depsgraph making
objects which are needed as dependency are not really evaluated.
Made it so setting frame via scene.frame_set() which check whether update
need to be flushed to an invisible objects and do this if so.
Not ideal solution but seems to be safest at this point.
Blender Internal shader nodes.
All nodes now use the `nodestack_get_vec` function to convert whatever
input values are connected to the desired type, instead of using the
stack values directly regardless of what they contain.
Note that this might break old node setups that have been adapted to the
faulty behavior, but as discussed in the bug report the 2.70 version is
the right time to fix this.
Issue was caused by curve orco calculation for rendering being freed
curve path and not calculating it back.
This left depsgraph in a state that it believed all the object data
is up to date but in fact some parts of data was freed by convert
blender.
Now made it so path is not being freed by render thread. This is
rather a workaround actually because ideally render thread need
to use copy-on-write here or at least use local cache here. But
current logic should be closer to what was happening in previous
release.
in Blender Internal renderer.
The BI renderer applies modifiers //after// changing the obmat of the
respective object (for the first instance it encounters). Before
rB6940bf0 the original obmat (omat) was stored inside dupli object data,
which was removed in favor of local omat variables due to hackishness
and redundancy. Problem with BI is that all the obmats have to be
overridden in relation to each other to produce the correct modifier
results (here: offset object for the array modifier).
The patch restores the old (messy) behavior for BI by first overriding
**all** the obmats at once from duplis, then creating render instances,
then cleaning up.
A better solution would be to avoid these modifier hacks in BI
altogether and properly evaluate them in the original object space, but
that requires far greater changes to the old code base, and is out of
scope for bugfixing.
Issue was caused by undefined object update order and in some
cases NULL pointer will be de-referenced.
Added on-demand curve path calculation, just the same creepy call
of BKE_displist_make_curveTypes(). This violates DAG and might
end up in a difficult to troubleshoot race condition if there'll
be some issues with how dependencies are calculated in DAG, but
this is the easiest and safest way to solve the bug at this stage,
BI.
The shaderfrom setting in the node editor only makes sense for "new"
shading nodes (cycles), otherwise it should be ignored and default to
SNODE_SHADER_OBJECT.
This commit implements dissolving of edges which were used
to triangulate non-flat faces. This slows things down a bit
(around 5% on heave mesh with all faces triangulated).
We could improve speed of dissolve a bit here (so not a bell
to add an option for triangulation yet).
Also fixed wrong edge origindex mapping.
reader.
To make a generic OSL shader connectable to other nodes, the parameters
must be declared via "input" and "output" child elements:
<osl_shader name="tex" src="./osl/stripes.osl">
<input name="Stripes" type="int" />
<output name="ColorOut" type="color" />
</osl_shader>
`name` must be the same as the OSL shader parameter name.
`type` must be one of float, int, color, vector, point, normal, closure,
string (matching cycles socket types)
Beyond this the OSL script nodes then work just like all other nodes.
OSL parameter sockets can be connected to other cycles nodes:
<connect from="checker color" to="tex Stripes" />
<connect from="tex ColorOut" to="floor_closure color" />
They can set default values for the input sockets by attributes of the
main node element:
<osl_shader name="tex" src="./osl/stripes.osl" Stripes="3" >
<input name="Stripes" type="int" />
<output name="ColorOut" type="color" />
</osl_shader>
This system of specifying custom attributes should probably be changed,
since it can easily create name conflicts and arbitrarily long elements.
But that is a different issue to be solved for all nodes in general.
- allow zooming when in ortho+user mode
- fix for using VIEW3D_OT_ndof_orbit_zoom when the camera is locked.
- fix for locked camera changing the dist value when used with ndof.
- de-duplicate ndof_pan_zoom and ndof_all operators
The file ##build_files/build_environment/install_deps.sh## contains the following line:
THREADS=`cat /proc/cpuinfo | grep processor | wc -l`
The command within the backticks is a [[ http://catb.org/jargon/html/U/UUOC.html | Useless Use Of Cat ]].
A more compact way of writing the same thing (saving two subprocesses) is
THREADS=`grep -c processor /proc/cpuinfo`
or (using POSIX-preferred command-substitution parentheses instead of backticks)
THREADS=$(grep -c processor /proc/cpuinfo)
But the most compact, and least Linux-specific, way is to use the ##nproc##(1) command from the [[ http://www.gnu.org/software/coreutils/manual/html_node/nproc-invocation.html | GNU coreutils package ]]:
THREADS=$(nproc)
Reviewers: sergey, mont29
Reviewed by: mont29
Differential Revision: https://developer.blender.org/D255
Basic idea is now to have the transformes bones keep "facing" the armature's Z axis, see comments in code for details.
That might not be ideal, but at least we now have humanly predictable and consistent results.
undo with a nicer enum.
This is more transparent about what happens with the user count.
Also added comments to make certain the meaning of these values is
easily understandable.
The node editor needs to "ensure real user" (id_us_ensure_real) on all
the node trees. This was done correctly for all open groups, but missing
for the base node tree. Function is a bit hidden behind anonymous
arguments in readfile.c ...
The issue was caused by the fact that sequencer used to cross-over effect
result with strips used for this effect, which is really stupid.
Now made it so strips which are used for effect inputs are not in the
render stack to be sure they would only be used by effect itself and
wouldn't be blended in any other way.
This mitigates T38505, where moving the mouse while using the wheel/keys
can exit the submenu.
So at least smaller unintentional movements won't be misinterpreted.
This can happen if no image buffers are used to define a sensible
resolution. Then the viewer will stiff create a float buffer in the
output imbuf, which defies the usual ibuf->rect_float check and leads
to invalid memory access. Float buffer should not be created in this
case.
Poll function tried to be smart and disallow having dependency cycles,
but it didn't work properly. Further, it still possible to re-shuffle
strips stack in the way which introduces dependency cycles.
So disabled this smartness in the poll function for now.
This allows to get the same "quickies" as in previous (2.69) code, (XYZ, -/, etc.), yet keeping nice non-conflicting new stuff like cursor navigation or copy/paste.
You can switch to full mode hitting '=', and back to simple mode hitting 'ctrl ='.
After some discussion it seems both are valid defaults but useful for
very different purposes.
- 'free' lets you explore the scene with full 6dof (like fly mode)
- 'orbit' is closer to typical mouse view orbit, constraining to orbiting about a central location.
This doesn't effect orbit/pan which are available with modifier keys.
For initial discussion see T38371
This commit organized panels for image editor to new tab categories dependent
on the image editor mode:
View Mode:
Tools - contains UV tools (currently only transform and UV Sculpting)
Scopes - contains scopes
Grease Pencil - contains Grease Pencil operators
Paint Mode:
Tools - contains brush options
Scopes - as above
Grease Pencil - as above
Mask Mode
Mask - contains mask tools
Scopes - as above
Grease Pencil - as above
Grease Pencil panel/tab now includes operators, not view options which have been
moved to the UI region on the right.
To make this work better, image editor toolbar now is of type TOOLS instead
of PREVIEW as was the case previously. A nice version patch makes sure all
works predictably, but opening newer files with older blender executables
could backfire.
This commit does not address which UV Tools will be included in the
Tools tab for the view mode, but does include some basic tools (transform)
and provides a class to inherit from to avoid conflicts with UV Sculpting.
Reviewers: brecht, dingto, sergey
Differential Revision: https://developer.blender.org/D315
* Switch to using Canon sensor type instead of camera models.
There are a ton of camera models that reuse the same sensor
so it's just confusing and forever incomplete to try and cover
all models.
* Delete all redundant Canon camera models..
* Leave only the Canon camera models that are special, that is
have non regular sensor sizes.
* Fix a few errors in some Canon models.
* Capitalize a few unrelated sensor presets for consistency.
* Added presets for GoPro and iPhone.
Their distortion models and shutter isn't really supported ATM
but tracker is robust enough to give reasonable tracking and
solving results.
Presets are from Daniel Salazar and Sebastian Koenig, thanks!
Differential Revision: https://developer.blender.org/D264
Some int/float conversion warnings were disabled by buildsystems but
re-enabled by BLI_winstuff.h, the warnigns relate to conversions not
considered issues on other systems so better just quiet them.
- avoid passing redundant operator name to keymap property set function.
- avoid double attr lookups when setting each property.
- handle exceptions on value type mismatch.
- avoid resource warning on failed load.
Previously 1/3 of the button was used to decrement, 1/3 to edit and 1/3 to
increment. However with the number text now right aligned this meant that
the increment area would overlap the number text, which is confusing. So it
was made to smaller to only cover the arrows.
It's not as easy to click but I don't know of a better solution with right
aligned number text.
nodes.
The Rotate node was calculating the center with a 1 pixel offset, which
effectively shifts the image by 1 pixel on one or both axis for
right-angle (90 degree) rotations.
Note that the wrapping feature for translate nodes can still produce
undesirable results for non-quadratic images. This is because of how
the resolution calculation works atm: the Rotate node will keep the
resolution of the input image, even if the resulting image is then
cropped or leaves empty margins. There is no easy way to fix that
without redesign.
It is possible that objects from dupli-groups depends on objects which are
not in the dupli-group. We do need this objects to be evaluated as well on
visible changes, so all dependencies for objects from the dupli-group are
met. Unfortunately, we don't have parent relations on this state, so we're
to use DFS over the whole DAG to gather such dependencies.
This is probably not so bad since visible update is called really rarely.
Reviewers: brecht
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D326
- moved assumed location of omp lib to blender libs
- prepared libiomp5 to link out of the box with cmake
- changed according in scons
- introduced a local var C_VENDOR, cause Apple clang 3.4 may not include omp support yet
- added a linklibs for msgfmt ( may not be needed for other than OSX )
* QE as up/down to Fly Navigator
* Alt as slow modifier to Fly Navigator
* Update on Walk header info
Note: I left the original RF/Ctrl shortcuts working for fly navigator
but they are no longer visible in the header info. So old users will
still be using what they were used to, while new ones get used to what
walk navigation is using.
The underlying cause for these issues is the insufficient sampling of
the bokeh image. For smaller blur radius there will be very few samples
taken, and with 1-pixel radius it boils down to just 4 samples:
2 on the left border (black), 1 in the center (black) and 1 at the top
border (blue) ...
For now have added the workarounds implemented in the CPU version of
that node, which hide these artifacts. Ultimately would be better to
have mipmap levels for the bokeh image input instead.
Was a shortcut conflict with unwrap. Use 'U' for unwrap now, so no conflict is here.
Preset still doesn't work tho, Cambo need to solve missing operators used here since
NDOF changes.
Non-camera objects are not supported as cameras, and trying to use them
just causes memory errors (some bad typecasts are involved). When converting
a Scene Actuator, only use the object as a camera if it is actually a camera.
Also, the logic brick UI now filters for camera objects, which should help
avoid confusion.
Cycles GPU Performance Regression
From my testing this (what i should have done in the first place) reduces the regression a lot.
Lets hope it is enough or we have to go back to busy waiting.
The test for a reflex angle used the vertex normal,
which was not a good test for a saddle point vertex.
Switched to using the face normal, if available, for that test.
Also added test for this in svn bevel_regression.blend.
This is because the sub-paths of the individual sockets are defined as
FILEPATH properties, which automatically adds the standard fileselect
operator button in the uiLists. However, that operator returns full
paths only, with no direct way to make them relative to the base path.
So for now i've turned the subpath properties into regular strings,
which removes these operator buttons but should at least avoid the
confusion.
Eventually new operators could be defined for this, which use the file-
select operator and then make the path relative.
Issue was caused by the change in FFmpeg options: some of them were
renamed, some moved to another class.
Made some tweaks to how options are passed to the FFmpeg which now
seems to be the same as ffmpeg.c.
Default Values is used on it.
The default for the socket value was always 0. Now the default value for
node socket default_value properties is chosen based on the socket
subtype to give some more sensible values.
NOTE: this may still be different from the values specified in the
socket templates! It is not possible to directly use these template
values, because all sockets share the same RNA types. At least it should
work reasonably well in the majority of cases now.
Issue was caused by missing GLOBALHEADER flag set for audio stream.
Also made it so frame is getting filled with default, that's what
happening in ffmpeg.c.
single terminator item.
Ideally no enum item function should return NULL, but since this is very
common and an intuitive mistake, better handle that case gracefully in
the RNA access function.
This makes it so NDOF without any modifiers pressed can pan and orbit
which means you have full 6dof, however if you only want to orbit+zoom
that works as before.
also change logic so rotation is disabled for axis views (orbit operator
will exit axis views still, as with mouse orbit)
nodes (Blur) causes crash due to chained read/write buffer operations.
The way read/write buffer operations are created for both the wrapped
translate node and then the "complex" blur node creates a chain of
buffers in the same ExecutionGroup. This leaves the later write buffer
operations without a proper "executor" group and fails on assert.
Solution for now is to check for existing output buffer operations like
it already happens for inputs. This is extremely ugly code, but should
become a lot more transparent after compositor cleanup ({D309}).
The color for frame labels was a mix of the text color (black by
default) and the node theme color with a factor of 0.8, which
coincidentally is the same as the node body color.
Changed it to 0.4, which is the same as the regular node labels use.
Disable transform and mask display when there's no active clip.
It's not a matter of returning fallback dimensions if there's no
slip, it's also matter of making it so stabilization and distortion
routines are aware of clip == NULL which is really crappy.
Also almost all the operators are disabled in clip editor without
active clip already anyway.
Also tweaked header UI a bit to not display mask stuff when there's
no active clip,
Allow duplicating splines from inactive layers. This way it's
more useful IMO than restricting duplication to an active layer.
TODO: What should be a behavior for clipboard (currently it copies
splines from an active layer only)?
Deadlock between python script evaluation thread and
drivers evaluation caused by scene update invoked from
ED_screen_set_scene().
Pretty much the same as setting scene current frame
with the same solution BPy_{BEGIN,END}_ALLOW_THREADS.
This adds an option in the Volume Sampling panel, which helps rendering lamps
inside or near volumes with less noise. It can also increase noise though and
needs improvements to support MIS and heterogeneous volumes, but since it's
useful in some cases already (especially world volumes) it's there now.
Based on the code in the old branch by Stuart, with modifications by Thomas
and Brecht.
Differential Revision: https://developer.blender.org/D291
updates.
This file crashes on loading with NULL pointer access to curve_cache:
{F77132}
The pdInitEffectors function was amalgamating the simple
collection of effector objects with an automatic precalculation for
curve guides and the like. This precalculation requires object data
that may not be available until the DAG has finished.
Since for DAG dependencies only the list of effectors is required,
added an argument to disable precalculation when collecting effectors.
when in ortho mode, zoom would dolly rather then zooming which
doesn't make much sense, now check ortho and zoom in that case
(nice behavior for locked quad view).
Goal of this commit is to support NGons for boolean modifier
(currently mesh is being tessellated before performing boolean
operation) and also solve the limitation of loosing edge custom
data layers after boolean operation is performed.
Main idea is to make it so boolean modifier uses Carve library
directly via it's C-API, avoiding BSP intermediate level which
was doubling amount of memory needed for the operation and which
also used quite reasonable amount of overhead time.
Perhaps memory usage and CPU usage are the same after all the
features are implemented but we've got support now:
- ORIGINDEX for all the geometry
- Interpolation of edge custom data (seams, crease)
- NGons support
Triangulation rule is changed now as well, so now non-flat
polygons are not being merged back after Carve work. This is
so because it's not so trivial to support for NGons and
having different behavior for quads and NGons is even more
creepy.
Reviewers: lukastoenne, campbellbarton
Differential Revision: https://developer.blender.org/D274
The blur operations were clamping the filter size to 1, which prevents
no-op blur nodes. Further any value < 1 would also be ignored and in
many combinations the filter scale setting ("Size") would only work in
integer steps.
Now most blur settings will work with smooth Size value scaling as well,
meaning you can choose a reasonably large filter size (e.g. 10) and then
use the Size factor to scale the actual blur radius smoothly.
Note that non-integer filter sizes also depend on the filter type
selected in the Blur node, e.g. "Flat" filtering will still ignore
smooth filter sizes. Gaussian filters work best for this purpose.
Updating rpms built from 'make package' has issues using the git hash since it isn't in a set order like svn commits, this lets it use the date to determine if the rpm is newer than the installed version.
Also updates the license text.
Those derived meshes (used in Edit mode) were using simple sum of neighbor poly normals to get vertex normals,
while everywhere else in Blender we use weighted sum of such poly normals.
Patch: D311
Reviewed and enhanced by Campbell, thanks!
- swap Z/Y axis so Z changes hue and X/Y change location.
this way you can use ndof like a joystick and Z rotation like a dial
for the hue.
- change sensitivity when shift is held to be 1/5th
This is really a workaround which brings back direct display
list creation from drawing code. This is rather nasty to do
but proper solution would require some major sequencer redesign.
Use Z axis for the edge direction for edges and vertex pairs.
Issue raised in T38592, now edge select and vert-pairs share logic
for calculating orientation and the active vertex determines direction.
The "Cast Shadows" worked as expected, but it can cause problem in some cases.
For example, when using strand render, we need disabling only buffer shadows,
but the previous changes made that impossible. "Cast Shadows" should be added
as a newly created option.
Since threaded object update we've disabled in-place
curve evaluation (in cases when applying curve modifier
with target curve non-evaluated yet).
This requires game engine to take care of DAG and object
evaluation (currently it's designed to export only objects
it able to render).
This workaround will make sure that curve_cache for curves
is up-to-date.
This allows us to make materials that don't cast ray shadows.
Turning off this property can reduce the rendering time slightly.
Note: RNA path is changed to "use_cast_shadows" as well. The older
path "use_cast_buffer_shadows" still can be used as its alias, but
it will be removed after updating some addons.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D272
Indirect and Direct samples can now be clamped individually. This way we can clamp the indirect samples (fireflies), while keeping the direct highlights.
Example render: http://www.pasteall.org/pic/show.php?id=66586
WARNING: This breaks backwards compatibility. If you had Clamping enabled in an old file, you must re-enable either Direct/Indirect clamping or both again.
Reviewed by: brecht
Differential Revision: https://developer.blender.org/D303
This adds tabs for Tools, Options, and Grease Pencil to Sculpt, Texture Paint, Vertex Paint, and Weight Paint. These tabs and their panel allocations were discussed in T38346.
Close T38346
While some of these are rather rough and crude, they should be sufficient to provide
some helpful hints to (new) users when mousing over the editor type selectors
(i.e. now we can see the names of the editors; without the descriptions, the enum
item names wouldn't get shown) and also when browsing the list.
This is useful now that we have the ability to display tooltips in menus, thanks
to Campbell's earlier commit (35f62bd)
FCurve Noise Modifer now has an extra float property which offsets the noise in time.
This is useful for creating follow through in procedurally animated noise.
For example, if you've used a noise modifier on a parent bone to add additional movement,
a quick and easy way to add overlapping motion is to create copies of that modifier on
its children, and then offset the time those curves play at. See this in action at:
http://youtu.be/Ph6fk_z_k3k
Reviewed By: Joshua Leung
On every redraw a single unopened dropdown boxe would translate
and convert every EnumPropertyItem into a string,
then decode every item, and search those items to find the name of the button to draw.
Replace this with a custom menu callback for RNA enums,
tooltips for enums now show too.
- use labels rather then menu items (items selected but did nothing)
- each report gets its own icon (icons besides first were ignored)
- use uiPupMenu rather then string based menu.
Removed Armature Sketch keymap, as the entries that were there
appear to have been moved to the Armature keymap.
Removed the Script keymap. The Script space is deprecated
and I could find no way that the keymap could be activated.
Bevel needs the vertex normals to be correct and they are not
normally recalculated after a previous modifier, like mirror.
Adding 'dependsOnNormals' -> returns true to modifier info
for Bevel fixes the problem.
Based on the patch from Sebastian Koenig, discussed with Jonathan Williamson
https://developer.blender.org/T38172
Also removed redundant modes from clip editor.
Reviewers: brecht, carter2422
Reviewed By: carter2422
CC: sebastian_k, carter2422
Differential Revision: https://developer.blender.org/D293
Ubuntu hack: add a --libyaml-cpp-ver parameter, to try to force installing a given version of this lib...
(only for DEB-like distro, I dare hope Trusty (!) is the only affected distro.
Not happy at all to have such things here, this is actually an Ubuntu bug, but will have to live with it I guess.
Description:
--------------------------
Use loose edges marked as seams as sewing springs.
Usage:
-------------------------
All this patch does is set the rest length to 0 and the stiffness to 1 for springs for loose edges marked as seams so that during the cloth simulation they will be brought together.
Example Video:
-------------------------
http://www.youtube.com/watch?v=-Y_bC0gjoM0
Original Patch by thesleepless (+ git patch by codemanx)
Thank you!
Z, Index, normal, UV and vector passes are only affected by surfaces with alpha
transparency equal to or higher than this threshold. With value 0.0 the first
surface hit will always write to these passes, regardless of transparency. With
higher values surfaces that are mostly transparent can be skipped until an opaque
surface is encountered.
This fixes the first case mentioned in the report. Has to do some ugly
DNA access to SpaceNode, unless we'd allow a bad level call there to do
it in ED_node ...
The second case has been fixed by @sergey in D274:
https://developer.blender.org/D274?vs=838&id=879#toc
So actually asan just did its job here, good to know!
This patch adds a network_error() function more alike how other devices handle error's
- it adds a check for errors on load_kernels to make sure we do not crash if rendering without a server.
- it uses the non throwing variation of boost::asio::read.
Reviewers: brecht
Reviewed By: brecht
CC: brecht
Differential Revision: https://developer.blender.org/D86
It was intended to work actually using particle cache's stack index
but this index might have been calculated incorrect in special case:
* With default cube scene, add particle system to the cube
* Add disk cache to the particle system
* Save file and reload it
* Add another particle system and enable disk cache
This would lead to two point caches with the same stack index of zero.
This happened because point cache indices list wasn't stored in the
.blend file so once you've reload your file blender doesn't know anything
about number or point caches used.
And what was even more confusing is that point cache indices list was
trying to be load from the file, but this failed because it wasn't in the
file.
This commit solves the root of the issue which is ability of producing
.blend file with two point caches using the same disk cache. This is
done by making it sure that point cache indices list is stored in the
.blend file. And also made it so disabling disk cache will tag it to
recalculate stack index.
Old broken files wouldn't magically start working, but fixing them is
rather simple manually by toggling Disk Cache option.
Reviewers: lukastoenne, brecht
CC: sergof
Differential Revision: https://developer.blender.org/D286
This in fact seems some kind of video driver limitation, because it worked fine
on windows but failed on linux here. The guess is that textures doesn't always
work on display lists, or we simply do have some wrong OGL context somewhere.
This is a workaround for until bigger viewport draw refactor is done (as Brecht
mentioned display lists are deprecated in new OGL anyway).
Root of the issue goes to the order of particle initialization which does
texture evaluation (which does depend on particle coordinate) and particle
birth coordinate calculation. So basically what happened is:
* Changing number of particles re-allocated all the particles,
which sets their coordinate to (0,0,0)
* Texture evaluation used this non-initialized coordinate
* Coordinates were calculated for particles
Reshuffled code a bit so now texture evaluation happens after particles.
coordinate calculation. Basically moved texture evaluation to particle
reset function. Reset happens after initialization anyway and it does
know particle coordinates. Also, if reset is being called without init
then it's also kind of logical to re-evaluate texture because particle
coordinates might change.
As discussed in T38340 the solution is to use the current scene from
context whenever feasible.
Composite does not use node->id at all now, the scene which owns the
compositing node tree is retrieved from context instead.
Defocus node->id is made editable by the user. By default it is not set,
which also will make it use the contextual scene and camera info.
The node->id pointer in Defocus is **not** cleared in older blend files.
This is done for backward compatibility: the node will then behave as
before in untouched scenes.
File Output nodes also don't store scene in node->id. This is only needed
when creating a new node for initializing the file format.
Reviewers: brecht, jbakker, mdewanchand
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D290
COLOR widgets never get created with UI_GRAD_V_ALT values. This is
intended for HSVCUBE color sliders only so reverting this to old value
before it was substituted with a define, which was -1.
There's one case where COLOR buttons get defined with a1 value different
than -1 which is in ui_def_but_rna. There the step value of the RNA
property is used. However, that meant that if step equaled to 9, which
is UI_GRAD_V_ALT value (unlikely) or if UI_GRAD_V_ALT was changed to 10
in future, or we added check for e.g. UI_GRAD_L_ALT which is 10 in HCL
branch, color widget would not get a color selection menu spawned.
Support for tagging polygon numbers when adding scanfill data,
saves having to calculate connectivity afterwards (which can take approx half overall scanfill time for complex curves).
The documentation for the -b/--background option incorrectly states that it
takes a <file>, however it can run also without a file with e.g. a script.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D250
The AVX kernel functions for reading image textures could be get used from non-AVX
kernels. These are C++ class methods and need to be marked for inlining, all other
functions are static so they don't leak into other kernels.
Issue was caused by NULL-pointer de-reference when post-processing
the frame without putting the frame to movie cache.
Differential Revision: https://developer.blender.org/D276
Issue was in fact in strip update code when transforming, in case we move both left and right
handles the strip is handled twice in the loop, but it was always updated at the end of the
first loop only...
On earlier versions there is no proper multi-monitor support, so we leave the
choice still. But for 10.9 this just interacts better with other elements like
spaces and the dock.
Crash was caused by missing curve_cache due to cyclic dependency.
This commit is rather a workaround, but we couldn't really guarantee
proper behavior in cases with cyclic dependency. So rather than calling
object update directly as it used to be let's don't perform curve
deform and be safe for threading instead.
This enables VEX-encoding in AVX kernel for windows msvc builds and gives 5-10% speedup for different scenes.
Reviewers: juicyfruit, dingto, brecht
Reviewed By: brecht
CC: brecht
Differential Revision: https://developer.blender.org/D284
Issue was caused by clearing dirty flags from element table in bmesh before
the data was fully ready.
This confused checks happening from threads and some threads did consider
element table is usable while other thread was still filling it in.
The Fix in 273 creates a backward incompatibility:
Collada files that have been created with an older Blender version
will contain the spotlight_size in Radians where Collada wants
this value to be in DEGREE.
This fix adds a check for the Blender Version that was used to create the
Collada file. If the Collada file was made by an older version of Blender
then the importer will assume that spotlight_size is specified in RADIANS.
Reviewers: campbellbarton, sauraedron
Reviewed By: sauraedron
CC: jesterking
Differential Revision: https://developer.blender.org/D279
Was a regression since avg_frame_rate changes.
Didn't find reliable way to get stream duration which will
work with both FFmpeg and Libav so added some freaking black
magic to distinguish one from another.
don't use extension checking preference for low level loading functions,
otherwise a user preference may impact a script loading a file for example.
looked into making this an argument and passing it in, but theres no
real advantage and its simpler just to remove.
Goes with companion files bevel_regression.blend
and mesh_ops_test.py just added to svn lib/tests/modeling.
Also changed the blender executable path for Apple case:
needed either 'Debug' or 'Release' in the path. I chose
'Debug', because not sure how to add otherwise. At any
rate, what was currently there didn't work at all.
This adds appropriate tabs, panels, etc for curves, surfaces, metaballs, text, and armatures. This brings it in line with the changes in rB0972c422c3ac4ed725414baa47838f833b3f4c90
This updates the object mode toolbar tabs to reflect the rationale from the Edit Mode tabs.
- Rename "Basic" to "Tools"
- Put "Transform" into it's own panel within Tools
- Also rename Edit Mode "Mesh Tools" to "Tools" for consistency and shorter names
- Ensure Tools tabs are in same location between Edit and Object modes
- Remove redundant "Transform" label from Transform panel
Added some relevant #defines copied from extern\libmv\third_party\msinttypes to BLI_sys_types.h so they can be used in blenlib. Made smallhash.c use the BLI_sys_types.h include rather than including <stdint.h> directly
The revision is concerned with Interface0DIterator and StrokeVertexIterator.
These iterators can be generated by Interface1D::vertices_end() and
Stroke::stroke_vertices_end(), respectively. These methods return an
iterator poinitng the next index of the last 0D element (i.e., iterator's is_end
property is true). When the iterators created in this way are used with
Python's iterator protocol (e.g., in a for-loop), iterations over 0D elements
are automatically performed in the reversed order. This functionality was
broken after recent revisions concerning Freestyle iterators.
Also made minor code cleanup (white space).
- use magic numbers based on uintptr max, not uint max, to avoid possible collisions with real pointer values on 64bit systems.
- comment BLI_smallhash_remove for now, its not used.
- added smallhash_val_is_used replacing ELEM() checks
- updated docs
- remove static array used only for copying (use alloca on resize)
- set SMSTACKSIZE to one of the values in 'hashsizes' since the full available size was never used.
- ensure ~1.5x as many buckets as entries, was 3x which caused malloc's quite early on.
Exported angles for spot size animation is in radians , however Collada expects it to be in degrees.
This patch is for fixing import and export, and also renaming a variable
Reviewers: gaiaclary
Reviewed By: gaiaclary
Differential Revision: https://developer.blender.org/D273
Issue partially caused by own errors (glicth in new BKE_boundbox_ray_hit_check() code causing segfault in volume snapping,
and we have to treat ortho and persp differently in case of face snapping, because in persp our ray_start might very well
already be *inside* the boundbox of the checked object), and partly due to the fact that ED_view3d_win_to_vector()
was returning wrong vector (negated one) for ortho views (see previous commit).
Found while working on snapping issues, confirmed using bisect tool: previous to this commit, inner/outer parts were swapped when switching from otho to persp!
The laplacian modifiers (smooth and deform) use the OpenNL library,
which is not threadsafe due to the use of a global context variable.
Ideally this would be changed so that an explicit context can be
created for every caller of the OpenNL functions, but since OpenNL's
most recent version is from 2010 this is unlikely to happen.
As a workaround for now just use a mutex to prevent conflicting OpenNL
calls. Eventually OpenNL can be replaced by eigen or ceres.
Also allowed me to fine another potential issue, hit.dist was no more initialized correctly - and I had forgotten to take into account Brecht's remark about normalize_v3() also returning the vector's previous length.
- BLI_smallhash_remove didnt decrement total entries.
- rename vars to match closer to ghash.
- smallhash_lookup returns NULL when no entry found.
- using a zero value key wasn't supported.
- no need to memset or calloc bucket arrays
- add asserts for unsupported conditions.
- added BLI_smallhash_lookup_p
Issue is caused by start point of ray used to detect faces under the mouse is set rather far away in ortho 3dviews.
The loss of precision on the ray location induced by this can lead to face snapping failures.
Solution is to do the raycasting with a temp start point, much closer to the object we check, and add back
to the found distance the diff to the real start point once detection is done (as we need all hit distances
from all tested objects to be relative to a common point!).
Note this commit only addresses the "face snapping on mesh" case, other kind of snapping do not seem to suffer
from this issue.
Reviewers: brecht, campbellbarton
Differential Revision: https://developer.blender.org/D268
This greatly improves the mesh edit tabs after a few weeks of feedback, making the tabs more practical with less need to switch tabs often. Based on discussion on IRC and T37568.
Reviewers: campbellbarton
CC: campbellbarton, brecht
Reviewers: campbellbarton
Reviewed By: campbellbarton
CC: campbellbarton, brecht
Differential Revision: https://developer.blender.org/D269
Added function called WM_set_locked_interface which does
two things:
- Prevents event queue from being handled, so no operators
(see below) or values are even possible to run or change.
This prevents any kind of "destructive" action performed
from user while rendering.
- Locks interface refresh for regions which does have lock
set to truth in their template. Currently it's just a 3D
viewport, but in the future more regions could be considered
unsafe, or we could want to lock different parts of
interface when doing different jobs.
This is needed because 3D viewport could be using or changing
the same data as renderer currently uses, leading to threading
conflict.
Notifiers are still allowed to handle, so render progress is
seen on the screen, but would need to doublecheck on this, in
terms some notifiers could be changing the data.
For now interface locking happens for render job only in case
"Lock Interface" checkbox is enabled.
Other tools like backing would also benefit of this option.
It is possible to mark operator as safe to be used in locked
interface mode by adding OPTYPE_ALLOW_LOCKED bit to operator
template flags.
This bit is completely handled by wm_evem_system, not
with operator run routines, so it's still possible to
run operators from drivers and handlers.
Currently allowed image editor navigation and zooming.
Reviewers: brecht, campbellbarton
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D142
- deduplicate timecode_simple_string from image.c
- replace V2D_UNIT_SECONDSSEQ with V2D_UNIT_SECONDS
- avoid possible buffer overflow bugs (sprintf -> BLI_snprintf)
- remove option not to use timecode and split into 2 functions
Patch D227 by Andrew Buttery with own refactoring.
The "pipe" case -- where two beveled edges are in line
and there is at least one more beveled edge -- needed better
handling when profile parameter = 1.0 (square outward).
This is my first stab at this and is based on this IRC converstation:
<mib2berlin> brecht: this is meaning as reminder only, I know you have other things to do > http://openvidia.sourceforge.net/index.php/Optimization_Notes#avoiding_busy_waits
<brecht> mib2berlin: thanks, bookmarked
only tested on Ubuntu 14.04 / cuda 5.0 but ill do some more testing tomorrow.
Also unsure about the placement and the lifetime of the stream and the event. But creating / deleting these seems to incur a non trivial cost.
Reviewers: brecht
Reviewed By: brecht
CC: mib2berlin, dingto
Differential Revision: https://developer.blender.org/D262
There is a hack for modal keymaps and double click events, which was (ab)using
the wmEvent.prevval variable to temporarily assign it KM_DBL_CLICK indicating
that the double click event was disabled. However the value of prevval can
actually be KM_DBL_CLICK in other circumstances too, which caused a key press
event to get converted to a double click event.
Additional bug fixes were also done along with the code updates:
* Fix for the use of old Interface1D.pointsBegin() and .pointsEnd() method names
in the definition of pyDensityAnisotropyF1D and pyViewMapGradientNormF1D.
* Fix for wrong data types (int instead of bool) for:
- pyChainSilhouetteGenericIterator constructor and its .orientation property in
modules/freestyle/chainingiterators.py.
- SpatialNoiseShader constructor in styles/external_contour_sketchy.py.
- ChainSilhouetteIterator constructor in styles/multiple_parameterization.py.
Both C- and Python-coded API components were rearranged into logical groups.
New Python modules are packaged as follows:
freestyle - Top-level package
freestyle.types - Classes for core data structues (e.g., view map)
freestyle.chainingiterators - Pre-defined chaining iterators
freestyle.functions - Pre-defined 0D and 1D functions
freestyle.predicates - Pre-defined 0D and 1D predicates
freestyle.shaders - Pre-defined stroke shaders
freestyle.utils - Utility functions
The Python modules are installed in scripts/freestyle/modules. Pre-defined
styles are installed in scripts/freestyle/styles.
To-do: update styles according to the new Freestyle API package structure.
First thing changed by this commit is making it so
Euclidean intersection takes track weight into account
when solving minimization problem. This behaves the
same exact way as it is for BA step.
Second thing is related on how average reprojection error
is being calculated. It didn't take track weight into
account which could confuse users. Now average reprojection
error will give the same result as intersection/BA uses
during minimization which gives much more predictable
behavior.
Differential Revision: https://developer.blender.org/D265
Using tracks with constant zero weight used to crash
keyframe selection since it was trying to use missing
parameter blocks for Jacobian evaluation,
Also fixed possible issues with wrong camera block being
marked as variable. This could technically happen when
having zero weighted tracks. Made it so all camera blocks
are marked as variable for now.
Switch the detector API to a single function which accepts
a float image and detector options. This makes usage of
feature detection more unified across different algorithms.
Options structure is pretty much straightforward and contains
detector to be used and all the detector-specific settings.
Also implemented Harris feature detection algorithm which
is not as fast as FAST one but is expected to detect more
robust feature points. It is also likely that less features
are detected, but better quality than quantity.
Blender will now use Harris detector by default, later we'll
remove FAST detector.
r_frame_rate is a guessed number defined as "the least common multiple of
all framerates in the stream". It has been deprecated and removed in new
Libav releases.
Use avg_frame_rate instead, which is the average framerate of the
stream.
CODEC_ID_* have been replaced with AV_CODEC_ID_* in new libavcodec
versions.
Update the code to use those new identifiers.
Added a compatibility code to ffmpeg_compat.h
Fixes some issues with NaN vertices in special cases.
Also adds edge interpolation routines which are currently
unused but which are requires to implement edge CD interpolation.
So now it's possible to copy-paste splines between layers.
Implementation is pretty much straightforward and duplicates
some logic which we've got in sequencer/tracking clipboards.
Will work on a common routine for clipboards later, for now
it's not so much crucial to have.
Updated CMakeLists.txt to account for boost hardcoded locations for VS 2013.
Reviewers: juicyfruit
Differential Revision: https://developer.blender.org/D261
Changed curve active point from pointer to index. Allows curve active point to be saved to file and retained between modes for free. Also some small optimisations by removing pointer look up code.
- Made active point access functions into BKE API calls.
- Fixes operators where curve de-selection resulted in unsel-active point.
- Split curve delete into 2 functions
* Very simple implementation, only allows for 1 output socket. As we haven't decided yet whether we keep the XML API, rather not spend more time on this now.
* To use an external osl shader, put the .osl file next to the xml file.
* Parameters: "output" is the output socket name, "output_type" the variable type (float, color and closure color are supported).
Example:
<osl_shader name="tex" src="ramp_closure.osl" output="Phong" output_type="closure color" />
<connect from="tex Phong" to="output surface" />
Factorized a bit the code here, think it's more readable now... No performance enhancement though.
Reviewed by: campbellbarton
Differential Revision: https://developer.blender.org/D259
* Change Info in header, put more important info to the left
* API: Move Camera width/height to camera, add some film properties
* Add ESC key to help menu
* EXR layers with names like 'Z' without any pass name were not loaded at all
and would break the Combined pass as well.
* EXR pass names longer than 16 characters where writing past the end of the
array and getting invalid names.
Until now, the "vertex group" limit method in the modifier
only worked for "vertex only" bevels. With this change,
edges with both ends in a vertex group will be beveled
in the non-"vertex only" case.
Also changed the test for being in a vertex group from
"any nonzero weight" to "weight >= 0.5". This is because
cascaded bevels on disjoint vertex groups did not give
disjoint bevels, because weight interpolation would give
non-zero weights to newly created vertices in earlier bevels.
Chose 0.5 because that won't result from interpolation, but
still allows some dilution (e.g., cascaded bevels on the
same vertex group).
libraries view not working.
This was disabled in the operator, there may have been a reason for this at
some point, but I can't see any reason to disallow it in the current code or
find a good reason why it was done in the commit logs.
The method for calculating points on the profile for non-circles
and non-lines meant that the segments making up an edge had
uneven widths.
Use a numeric search technique to find superellipse evaluation
points that lead to equal-sized chords across the profile.
Also calculate the actual profile points sooner, so that they
don't have to be recalculated again and again.
This also sets up for a possible later feature of arbitrary
profile shapes, set by user.
This is very basic for now, but can be extended with more info (available devices for example) later.
Thanks to Bastien and Sergey for some help with the glRect coordinates stuff.
Previously the divider was using the region background and the region text, leading to bad outlines with custom themes. This makes the outline behave more expectedly and allows better control.
This is rather a workaround which only works because curve evaluation is only called
for a temporary object. Not a big deal if we'll skip path creation for such objects.
Still would need to think of general solution.
For example for vector displacement, you may have an EXR texture that has
negative colors values. Blender clamps these by default, now the Colors panel
for textures has a Clamp option to disable this clamping.
This option affects all texture types and is enabled by default, you need
to disable it if you want negative values to have an influence.
Patch by Fredrik Hansson with modifications by me.
it is possible that different windows shares scene but displays different
layers. And it's also possible that different areas in the same window will
show different layers.
First case was violated in `dag_current_scene_layers()` which only checked
scene layers only once and if multiple windows shares the same scene only
one window was handled. Now made it so layers from all windows will be
squashed together into a single `DagSceneLayer`. This mainly solves issue
with `DAG_on_visible_update()` which didn't work reliable with multiple
open windows.
Second case required call of `DAG_on_visible_update()` when changing space
are type.
This commit slows things a bit actually because `dag_current_scene_layers()`
is actually called on every main WM loop iteration. It is possible to speed
some logic up perhaps. Not sure it's so much critical to do now because there
are unlikely to be more than few windows open anyway.
Will rather think of skipping all that flushing things if no objects are
tagged for update actually.
- Made them receive number of channels rather than number of planes.
This matches to how ImBuf structure stored planes and channels.
- IMB_premultiply_rect_float() was called with channels passed instead
of planes already :S.
Issue is caused by the race condition between getting custom data layers
from target's derived mesh (for vertices and faces) and releasing this
derived mesh from other threads.
When one releases the derived mesh it'll free temporary data from it,
and it'll also update data layers mapping.
General rule for threading is that no one is ever allowed to modify
data he doesn't own. This means that no temp layers are to be allocated
in derived mesh and making it so `CustomData_free_temporary()` doesn't
update mapping if nothing was freed will solve the race condition.
It is still possible to do other improvements, namely detect which
additional data/layers are to be present in derived mesh and create
it as a part of `object_handle_update()`, but this is to be solved
separately.
This was storing the original object matrix, which builds on the
assumption that obmat is modified during dupli construction, which is a
bad hack.
Now the obmats are still modified, but this only happens outside of the
dupli system itself and the original ("omat") is stored as local
variables in the same place where the obmat manipulation takes place.
This is easier to follow and avoids hidden hacks as much as possible.
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D254
patch D234 from Jonathan Williamson with edits
- de-duplicate rna_def_userdef_theme_space_gradient and rna_def_userdef_theme_space_generic
- ui_theme_init_new_do now always sets theme settings (no need to test),
used by bpy.ops.ui.reset_default_theme()
On Windows we can only do mmap memory allocation up to 4 GB, which causes a
crash when doing very large renders on 64 bit systems with a lot of memory.
As far as I can tell the reason to use mmap is to get around address space
limitation on some 32 bit operating systems, and I can't see a reason to use
it on 64 bit. For the original explanation see here:
http://orange.blender.org/blog/stupid-memory-problems
Fixes T37841.
Own bug from rBc691551249f3. Now at least I understand why `test_index_face()` is needed for tessellated quads!
Added a bunch of comments to explain the issue, as it's far from an obvious one...
We loose some performances, but it's still much quicker than org code.
Issue was caused by curve object really scale up. It was
caused by 677f519 to make scaled down curves work fine.
After some tweaks to epsilon value scaled up curves seems
to work as well.
Issue was caused by wrong PTS calculation. This commit
makes this calculation closer to what's happening in
FFmpeg itself.
Seems everything is working now including newer FFmpeg,
but there's one thing which still doesn't work: writing
avi files with h264 codec and Vorbis audio doesn't play
correct in mplayer here. But didn't manage to get this
working even using FFmpeg CLI, so this might be just a
bug in FFmpeg/mplayer. Since this file works fine in
blender just fine wouldn't consider this is crucial thing
to look into at this moment.
(@plasmasolutions): When the particle emitter is parented to a fast
moving object, the emission locations will not be interpolated over
subframes. This works if the particle emitter is animated itself.
Particle system evaluates the emitter location for each subframe, but
has to do this for the parent objects as well to get reliable results.
The area-of-interest calculation for that node didn't work reliably.
It tries to estimate the distorted rectangular area based on min/max
distortion and dispersion values, but this fails in some cases and
leaves uninitialized buffer chunks. So now simply use the full input
rect as the area, even though it may not be as efficient - at least it
works ...
Also cleaned up the code somewhat to make it understandable, using
separate functions for common stuff instead of cryptic walls of math.
FileGlobal structure in write_global() didn't initialize
the while build_hash property and only placed there null
terminated meaningful content, leaving trailing uninitialized.
This made valgrind complain about using uninitialized memory
later in undo stack.
Evaluating the animation is causing the object to get tagged as changed, but in
this case it's not a permanent change so no one should be notified. Also found
a case where the persistent ID for duplis wasn't unique, fixed that as well.
The main idea is to store (during tessellation) or recreate (during tessdata update) a tessfaceverts-to-loops mapping, and then update all tessdata
in one pass, instead of calling `BKE_mesh_loops_to_mface_corners` repeatedly for all tfaces!
Differential Revision: https://developer.blender.org/D226
Reviewed by Campbell, thanks a lot!
Ref: http://en.wikipedia.org/wiki/XOR_swap_algorithm, modern compilers/CPUs are much more efficient with "naive" algo than XOR one.
Doubled check, for me in an optimized build, XOR is several times slower than naive algo.
It was using the modified mesh for the vertex coordinates, and the unmodified mesh
for the vertex weights, which can't work and crashed in some cases, now it used the
modified mesh for both.
Perhaps an option should be added to indicate if you want to transfer from the
unmodified mesh or transfer to the modified mesh, but I think this fix makes it
works as intended when this was implemented.
With new subdivision method for making the vertex mesh, we always
subdivide the edges between the new vertices around a vertex
(we used to not subdivide edges between two non-beveled edges).
This fixes a bug related to this change.
The string comparison was in lower case, so the same strings with different case
were considered the same which can make qsort give different results on each
sort since it's not a stable sort. Now take case into account in comparison.
This is a first step toward improving our dupli system. It implements a more
generic way of treating the various methods of dupli generation by adding a few
structs:
* DupliContext holds a number of arguments commonly used in the recursive dupli functions and defines a recursion state for generating sub-duplis (nested groups). It also helps to prevent bloated argument lists.
* DupliGenerator is a type struct that unifies the different dupli creation methods (groups, frames, verts, text chars, faces, particles). (As with context there should be no overhead from pointer indirection because everything can still be inlined inside anim.c)
Beside making the code more easily understandable this implementation should
also help to avoid weird side effects from custom matrix hacks by defining
clearly what a generator does. The DupliContext is deliberately made const, so a
generator can not simply add hidden matrix or flag modifications that are hard
to track down.
The result container for the generated duplis is stored in the context instead
of being passed explicitly. This means the generators are oblivious to the
storage of duplis, all they need to do is call the make_dupli function. This
will allow us to implement more efficient ways of storing DupliObject instances,
such as MemPools or batches. These can be implemented alongside the current
ListBase so we can improve dupli bottlenecks without having to replace each and
every dupli use case at once.
Differential Revision: https://developer.blender.org/D189
The issue was caused by missing object update for the curve object
before going to new screen. And that new screen had different visible
layers due to disabled lock_camera_and_layers option.
Solved by calling DAG_on_visible_update() in ED_screen_set() even
in cases scene doesn't change. This ensures all the objects which
weren't visible before are correctly updated.
Code here was using pchan->chan_mat, which for some reason does not work,
we rather have to use pchan->pose_mat, and convert it again from
pose space to bone space, as done by visual keyframing.
On Linux/Mac OS X, simply type "make cycles" inside the Blender source directory, to get a standalone build of the engine.
Reviewed by: Brecht
Differential Revision: https://developer.blender.org/D228
Allow loading of image sequences in addition to single images in the open image operator
I solved it by adding the possibility to load an image sequence in the Load Image Operator.
The image user is passed in the operator customdata now as well, best solution for now, but general handling of image user still a bit weak. The offset and length of the image sequence is now calculated in the image open operator by storing all found frames in a list and sorting this list.
Reviewed By: campbellbarton, brecht, lukastoenne
Differential Revision: https://developer.blender.org/D209
LSCM auto pinning where it picks two vertices on opposite sides of the outer
UV island boundary would in some cases give the same 3D coordinate, which
causes UV unwrap to give poor results.
This code has been broken for a few years and no one noticed, it's also less
useful now that we support PSD image loading ourselves which was the original
motivation to have this.
Vertex bevel was completely messed up after last change,
and this fixes.
'Weld' bevels, where there are exactly two beveled edges
meeting at a vertex, now look better in cases where one
of the beveled edges is on a flat plane.
It is possible that object in .blend file will have non-zero
recalc flags. Clear them on load in order to prevent possible
unneeded recalc after the load.
Differential Revision: https://developer.blender.org/D224
This is because of the animdata cleanup in rBd2e55cb. This works ok in
general, but causes issues with the localized node trees used for compo/
shader/texture previews. These localized trees share the same default
action as their original trees, and then remove fcurves when removing
muted nodes (which should affect the localized tree only).
node_free_node_ex now has an argument for disabling animdata cleanup,
which is also not necessary when freeing the whole node tree (because
animdata is freed in advance anyway). In addition to that it also checks
the NTREE_IS_LOCALIZED flag to prevent freeing of fcurves in the action.
- move cache reset into view2d and comment about the rationale for resetting cache there.
- missed other places where view2d zoom level can change (smoothview, borderzoom, home and file reload)
- comment about DPI wasn't correct.
Code is not equivalent in min/max part (SSE works with NaNs differently), this results in black dots with cardinal_curve hair.
This reverts commit b886c26d1f.
Issue is not noticeable with default font, but i18n one can use more than 500Mo of ram when cached in undreds of different sizes.
Campbell had already done most of the work in rBa780e7f3f09f and rB6b283f116829, just had to call BLF_cache_clear() in _exit funcs of VIEW2D_zoom & co operators.
Increase the maximum allowed amount of points in a spline from currently 32,767 (short) to 2,147,483,647 (int).
Change variables that get assigned the value from pntsu/pntsv to int type all over the codebase.
Change function parameters that previously passed the count as short to int type as well.
(because https://developer.blender.org/T38191)
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D212
During drag the H key can be used to toggle the hide flag of the selected nodes.
This makes it easier to 'attach' nodes to available links in narrow places.
Adds some padding to the left of the icon, adds more space around the separator
line and make it more visible, and add some spacing at the top and bottom of
the menu. Ref T37794
Reviewed By: dingto, billrey
Differential Revision: https://developer.blender.org/D223
It was an issue with early object update check which was
screwing up for second scene because of first one reset
ID recalc flags.
From the comment in the code about this:
We need to check all visible scenes, otherwise resetting
OB_ID changed flag will only work fine for first scene of
multiple visible and all the rest will skip update.
This could also lead to wrong behavior scene update handlers
because of missing ID datablock changed flags.
This is a bit of a bummer to allocate list here, but likely
it wouldn't become too much bad because it only happens when
objects were actually changed.
A logic error meant that the wrong "previous rotation" values were being used
when decomposing the rigidbody results back to transform channels. Instead of
using the previous values for the object in question, it was actually using
the rotation value of the previous object that was evaluated.
This commit introduces the ability to make the Build Modifier operate in reverse,
essentially allowing it to be used as a "deconstruction" effect.
(See D219 for more details about use cases for this)
* Version patching fixes for theme settings
* Added missing support for NLA (needed for the keyframes drawn in the action lines)
* Fix for a lack of contrast between selected and unselected extreme keyframe type
(restoring it back to the pre-patch color scheme)
* Fix for keyframes on protected channels not being drawn with partial opacity
This patch makes it possible to customise the colours used for the different
keyframe types (Keyframe, Breakdown, Extreme, Jitter) and the border colours
(normal and selected).
Reviewed by: Joshua Leung
This should be pretty rare, the shader in question had many parallel node links
because of copying the nodes many times, which is inefficient to run anyway.
Each dpi value stores its own set of font sizes, so while dragging the
dpi value would collect many sizes and never free.
Also change how BLF_cache_clear works,
it was freeing memory but not the OpenGL textures.
Now just free all the cache and GL textures and let drawing allocate them again as needed.
* AVX is available on Intel Sandy Bridge and newer and AMD Bulldozer and newer.
* We don't use dedicated AVX intrinsics yet, but gcc auto vectorization gives a 3% performance improvement for Caminandes. Tested on an i5-3570, Linux x64.
* No change for Windows yet, MSVC 2008 does not support AVX.
Reviewed by: brecht
Differential Revision: https://developer.blender.org/D216
Issue was caused by solidify modifier using original vertices bitfield
to store tags. This isn't thread-safe obviously. Now use bitmap to store
needed tags.
Reviewed by Campbell, thanks!
Issue was caused by some objects being in bMain and tagged
for update but not being in the DAG. This means objects
wouldn't be updated and their recalc flag remains untouched
triggering threaded for the next frame.
Solved by tweaking POST_UPDATE_HANDLER_WORKAROUND in the way
that it checks objects' recalc flags from the DAG, not from
the bMain. This will work a bit longer since DAG stored more
nodes than objects in the scene, but this code only runs in
cases when there're some objects tagged for update, which
keeps overall CPU usage on such a workaround pretty low.
Now CPU usage on 11a_comp scene from project Pampa went down
from ~15% down to ~5% (2,69 release uses ~%7).
Pointed by Thomas Dinges in IRC.
This is to try to avoid some confusion now that the separate text fields are gone.
Reviewed By: billrey, brecht
Differential Revision: https://developer.blender.org/D217
Previously this only worked for some datablocks relevant to rendering, now it
can be used to detect if any type of datablock was added or removed (but not
yet to detect if it was modified, we need many more depsgraph tags for that).
Most of the changes are some function parameter changes, the important parts
are the DAG_id_type_tag calls.
Reviewed By: sergey, brecht
Differential Revision: https://developer.blender.org/D195
Because of the issue of how simplification works, there could
be more flags in on->recalc bitfield which are ignored by object
anyway.
Would save some update calls after simplificating the scene.
It was actually rather simple to make sure depsgraph is up to date
before calling BKE_object_handle_update() in the transform code by
just moving DAG_scene_relations_update() to the beginning of the
function.
Wouldn't expect any side effects for other cases since DAG is not
used bu the code between new location of the call and old one.
Issue is caused by the evaluation flags getter called with
NULL depsgraph. It happens on direct object update from the
transform code after duplicating the curve.
Proper solution is probably to make sure depsgraph is rebuild
after duplication, but for now it's better to prevent crashes.
Issue was caused by evaluation flags getter function polluting
the DAG. Need to use dag_find_node() instead.
Still need to doublecheck exporting objects with curve deform
works properly. On the first thought it should, but might be
wrong again.
We had several reports where users would create rigid bodies by accident
and then wonder why transformations behave differently.
Since these actions aren't used that often, just remove the shortcuts.
Weird library names having version suffix (e.g. libIex-2_1.so) are used only
when the OpenEXR was built with CMake. Building the library with the Autotools
doesn't add the version suffix.
This makes the tabs slightly larger, increases the text size slightly, and adjusts the colors a bit to improve overall contrast. It also makes inactive tabs draw with a full tab shape, improving the readability at a glance, particularly when tabs are very small.
The users most affected will be those on smaller displays, where previously tabs were very difficult to read.
Reviewers: @billrey, @campbellbarton
Differential Revision: https://developer.blender.org/D170
This adds an ImageUser to such empties with all the typical settings.
Reviewed By: brecht, campbellbarton
Differential Revision: https://developer.blender.org/D108
With edits by Campbell, thanks!
Looks like in some cases (driver dependent?), `XDeviceMotionEvent` get generated with only part of expected data
(e.g. only x coordinate, only pressure, etc.), data which did not change since last event being NULL.
We know which data to actually handle with `XDeviceMotionEvent.first_axis` and `XDeviceMotionEvent.axes_count` values.
Reviewed by: campbellbarton
Differential Revision: https://developer.blender.org/D208
this allows for updating icons without committing a new PNG each time
(which is inefficient with git). The data files are converted into a
PNG at builds time and used just as they were before.
It was a missing fcurve evaluation in scene update function which lead to
materials only being updated on frame change.
Added the same exception as we've got for the scene animation. It only
runs when there're materials tagged for the update, so wouldn't expect
speed regressions or so.
Issue was caused by cache limitor removing viewer image buffer
from the memory during compositing. Now made it so all viewer
images are persistent in the memory.
This solves the crash mentioned above and also makes it so
render/compo results are never lost.
Further tweaks are possible, but pretty much happy now, at
least no stoppers for work are there.
Same as mballs and curves it was only to work around DAG stupidness
which hopefully was fixed in one of the previous commits.
From quick tests everything works fine, if something is broken now
poke me to find a proper solution.
There were couple of reasons why it wasn't safe for usage from
multiple threads.
First of all, it was trying to cache BVH in derived mesh, which
wasn't safe because multiple threads might have requested BVH
tree and simultaneous reading and writing to the cache became a
big headache.
Solved this with RW lock so now access to BVH cache is safe.
Another issue is causes by the fact that it's not guaranteed
DM to have vert/edge/face arrays pre-allocated and when one
was calling functions like getVertDataArray() array could
have been allocated and marked as temporary. This is REALLY
bad, because NO ONE is ever allowed to modify data which
doesn't belong to him. This lead to situations when multiple
threads were using BVH tree and they run into race condition
with this temporary allocated arrays.
Now bvhtree owns allocated arrays and keeps track of them, so
no race condition happens with temporary data stored in the
derived mesh. This solved threading issues and likely wouldn't
introduce noticeable slowdown. Even when DM was keeping track
of this arrays, they were re-allocated on every BVH creation
anyway, because those arrays were temporary and were freed
with dm->release() call.
We might re-consider this a bit and make it so BVH trees are
allocated when DM itself is being allocated based on the DAG
layout, but that i'd consider an optimization and not something
we need to do 1st priority.
Fixes crash happening with 05_4g_track.blend from Mango after
the threaded object update landed to master.
Graph traversal which is based on counting parents which are still
to be updated fails in cases there are cycles in the graph.
If there are cyclic dependencies in the scene all the objects from
the cycles will be updated in a single thread now one by one. This
makes blender behave the same way as it was before multi-threaded
DAG landed to master.
This needed to tweak depsgraph a bit so now dag_check_cycle() sets
is_acyclic field of DAG forest if there are cycles in the graph.
TODO: It might be possible to save some time on evaluation when
all the tagged objects were updated in multi-threaded DAG
traversal.
This is a regression since threaded dependency graph landed to master.
Root of the issue goes to the loads of graph preparation being done
even if there's nothing to be updated.
The idea of this change is to use ID type recalc bits to determine
whether there're objects to be updated. Generally speaking, we now
check object and object data datablocks with DAG_id_type_tagged()
and if there's no such IDs tagged we skip the whole task pool creation
and so,
The only difficult aspect was that in some circumstances it was possible
that there are tagged objects but nothing in ID recalc bit fields.
There were several different circumstances when it was possible:
* When one assigns object->recalc flag directly DAG flush didn't
set corresponding bits to ID recalc bits. Partially it is fixed
by making it so flush will set bitfield, but also for object
types there's no reason to assign recalc flag directly. Using
generic DAG_id_type_tag works almost the same fast as direct
assignment, ensures all the bitflags are set properly and for the
long run it seems it's what we would actually want to.
* DAG_on_visible_update() didn't set recalc bits at all.
* Some areas were checking for object->recalc != 0, however it is was
possible that object recalc flag contains PSYS_RECALC_CHILD which
was never cleaned from there.
No idea why would we need to assign such a flag when enabling
scene simplification, this is to be investigated separately.
* It is possible that scene_update_post and frame_update_post handlers
will modify objects. The issue is that DAG_ids_clear_recalc is called
just after callbacks, which leaves objects with recalc flags but no
corresponding bit in ID recalc bitfield. This leads to some kind of
regression when using ID type tag fields to check whether there objects
to be updated internally comparing threaded DAG with legacy one.
For now let's have a workaround which will preserve tag for ID_OB
if there're objects with OB_RECALC_ALL bits. This keeps behavior
unchanged comparing with 2.69 release.
First of all, it was needed to have that set scenes fix which
was done recently so curve is being evaluated properly on file
load.
And last but not least, also needed to tag DAG node to evaluate
path regardless to curve datablock settings so curve length is
always known.
This solves threading conflict which happens when having
multiple objects using Curve Deform modifier with the same
curve datablock. This conflict was caused by the fact that
curve_deform_verts() used to temporary override curve's
flags to make it path is there.
Actually, it was setting CU_FOLLOW flag temporary which
was only used where_on_path() (only in terms that this
temporary assignment only affected this function) but it
is now commented out for a while, so no reason to set
this flag temporary, If it's ever to be done, we'll need
to pass flags as an additional function argument.
For the path creation i've extended DegNode structure
which now holds extra bits which indicates what additional
data depending on the graph topology is to be evaluated.
Currently this is only used to indicate that curve needs
path to be evaluated regardless to cu->flag state. This
is so Curve Deform modifier is always happy.
In the future this flag might also be used to indicate
whether bmesh verts are to update (see recent commit to
3-vertex parent crash fix) or to indicate that the object
is the motherball etc.
This goes back to ancient era again and such a call isn't
safe for threading and really DAG is to make it sure display
list for dependencies is always there.
BKE_curve_bevel_make() is only used from object_handle_update()
friends and never called directly. This means if there's no
display list ready for the bevel object it's something wrong
happened with DAG.
In fact, this check goes back to ancient era and from tests
it appears this check is no longer needed.
It was some kind of workaround for DAG glitch in 2009
(commit hash 8c5c7ebb0) and according to the comment
was needed to make select outline show immediately.
After some tests it appears DAG behaves almost fine now
(just needed to make it so layer is flushed properly to
the set scene) and no reason to have rather confusing
call in the code.
We do keep it when typing an new directory name in the directory text field.
Reviewed By: brecht, billrey
Differential Revision: https://developer.blender.org/D188
It is better to keep the profile as it is perpedicular
to the edge, and then project it onto a given plane
at the corners. Also fixed the interpolation to a
different number of segments when the profile is not
round.
Issue was in GP draw code: line thickness was not initiated properly (and a null one makes OGL draw a unity-width line).
Also tweaked threshold (when to start a new, width-different OGL linestrip), to make it inversaly proportional to thickness
(so that now, it's 0.2 for a thickness of 1, 0.1 for a thickness of 2, etc.), avoids too big steps when using large thickness.
Gives up to 15% speedup scenes with voronoi-based textures (up to 25% with volumes) on Haswell. The performance change for other CPUs is much smaller: 1-2%.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D203
This simply mimics code used for loopnormals, to enable py scripts to generate and access (temporary)
a tangent 3D vector and bitangent sign for each loop. Together with the split normals, this allow
to recreate a complete tangent space for normal mapping (bitangent = bitangent_sign * cross(normal, tangent)).
Expects all faces to be tri or quads.
Reviewed By: Brecht, campbellbarton
Differential Revision: https://developer.blender.org/D185
group default values.
This can happen when using value sliders for node group input values.
The localized copies were setting the "interface_type" runtime pointer
of the original tree to NULL instead of the new tree (which is created
on-the-fly in general). This type is used in RNA update functions
however, the original tree DNA should not be modified there.
This was (more or less) OK with hand-drawn strokes, as the number of points made it nearly unoticable, but broke completely with line and poly strokes!
Did this when I implemented linked curve feature because it was easier! Now, convert code always adds a heading and trailing point to the curve,
to get initial/final zero radius. Adds even more complexity to those functions... :/
DAG node tagging was rather an experiment to make derived render working.
However, it ended up in a whole can of worms and need to be re-considered.
It is likely that regular object update tagging and scene update routines
are to be used for this.
Meanwhile no need to keep extra field in dag node. Would save us the whole
byte of the struct which we can use for other purposes meanwhile.
Issue was introduced in a2bf25e and was caused by
do_makeDispListCurveTypes() no longer placing nurbs
to cu->nurb list.
Such an operation isn't thread-safe and proper solution
would require having granular update. For until them
just make object conversion take care of filling cu->nurb
in with splines from font.
When trying to scale bones in EditMode and the bones were drawn using envelope display mode,
this resulted in the joint radii (i.e. the inner part of envelopes) being adjusted instead.
It turns out that this was due to an old hack that was put in place back in 2.4x (see the
tracker logs for full details of the problem here).
This commit introduces the following fixes:
1) Removed the old hack. Scaling (S) works normally now.
2) Ctrl-Alt-S (i.e. "Scale Envelopes/BBones") is as-is.
That is, it is used to adjust the size falloff-region around a bone
(i.e. the "dist" property)
3) Added Alt-S hotkey in EditMode for armatures for adjusting the radii of bones.
This change just means that the "TFM_BONE_ENVELOPE" mode is now able to be
accessed from the UI as a tool on its own right (instead of being accessible
via the old undocumented hack). This tool adjusts the radii of the bone joints,
which define the actual full-influence region of the envelopes.
Issue is causes by vertex parent modifies original BMesh from
a multiple threads. Ideally this is to be done as a separate
update callback for mesh datablock, but it's not so much simple
now (would need to do some re-arranges to DAG which might conflict
with the work from Ali or will double amount of work we did).
So for now use simple solution with mutex lock.
Based on the patch from Campbell Barton with some fixes to make
changes really thread-safe.
Differential Revision: https://developer.blender.org/D168
to previous sculpt fix (OpenGL access from thread with no context
bound). The fact that this has gone unnoticed so far means that people
are dyntopoing like crazy these days.
It doesn't make any sense anymore with the current depsgraph and probably was
not useful for a long time, just a leftover from the pre 2.04 game engine.
Due to float precision issues it was basically random which of the two was used,
now it's slightly biased towards 128, which is the convention for flat colors.
The small difference between 127 and 128 could give problems with sharp glossy
shaders where it would be visible as seams.
Use manual ternary operation widening in grad(). Without it nvcc 5.5 produces multiple branch splits with very big branches (because of inlining). This solves 19% performance regression for BMW1M-MikePan.blend.
Also remove one redundant instruction in perlin SSE (when h == 12 or h == 14, then h is always >= 4).
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D190
When yiz insert a keyframe (on a location keying set f.e.) in the 3D-view and you got an action editor open then the action datablock is not changed (displaying only the button "new") until you hover the action editor with the mouse.
Added a handler in the action_header_area_listener to handle this case
If non-obvious, some technical note about what the cause of the bug was and
how it was solved.
Expose float precision helper in UI_interface.h API, so that numinput can use this helper as numbuttons already do.
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D186
Parameter controls concavity / convexity.
<.25 means: concave inward
.25 means: straight slanted
>.25 means: concave outward
.5 means: circular (the default)
1 means: straight along original sides
For now, there is a hard lower limit of .15
because more work is needed to get decent
results in the range below that.
The profile is actually a superellipse, and the
parameter is 1/4 of the exponent in the implicit equation
for a superellipse, except at the extreme values of 0 and 1.
to exist and accessed without prior NULL checks (with exception of
get_next_frame and get_movie_path). The callbacks are not reliably
initialized however if none of the video formats is enabled (AVI being
the default). Added stub functions now that ensure access to
bMovieHandle callbacks is safe and doesn't crash.
This is done by adding a Volume Scatter node. In many cases you will want to
add together a Volume Absorption and Volume Scatter node with the same color
and density to get the expected results.
This should work with branched path tracing, mixing closures, overlapping
volumes, etc. However there's still various optimizations needed for sampling.
The main missing thing from the volume branch is the equiangular sampling for
homogeneous volumes.
The heterogeneous scattering code was arranged such that we can use a single
stratified random number for distance sampling, which gives less noise than
pseudo random numbers for each step. For volumes where the color is textured
there still seems to be something off, needs to be investigated.
* Now only middle-clip when both left and right parts of remaining strings are at least two times icon-width (thus most short labels will never be middle-clipped).
* Add a trailing ellipsis to right-clipped strings (unless it would take more than 20% of whole available width, in this case just clip chars, to keep as much data visible as possible!).
The sanity check for copying socket default value was using the socket
idname for type compatibility checks, which is too strict. Subtypes
would not be recognized as copyable, but since only the plain data type
is needed below this is all that needs to be checked.
That alone would not cause crash (just missing default value copy),
but the tree view template was messing with the default_value DNA
directly by freeing it in advance, which is not necessary and should
be left to blenkernel when freeing the node. Otherwise this would
leave the node invalid without a default_value if the copy function
bails out.
[ foo = bar ] has been the traditional compare operator, using == is supported by some shells, but not all. The latter offers no advantage over the former.
Differential Revision: https://developer.blender.org/D179
In separating out the adj mesh change from a profile change,
I'd forgotten that some profiles need to be flat (in particular,
for vertex-only bevels). This fixes that.
This gets rid of a hacky way of setting the mesh
coordinates at corners, which created overlaps
in some cases.
Also, special case a cube-like corner with all
edges beveled, to snap to a sphere, giving more
expected curvature in that case.
Originally by Dan Eicher, with my own fixes and adjustments (see patch page for details).
For details there are unit tests and api example usage.
doc/python_api/sphinx-in-tmp/menu_id.png
- Add paste from system clipboard which behaves like paste from file.
- Paste from file now replaces the selection rather then just adding to the end.
- Move paste operations into the 'Edit' menu.
- Added generic paste functions: font_paste_wchar, font_paste_utf8.
- Fix paste max length check not taking the selection length into account.
Issue was caused by BKE_vfont_to_curve() modifying curve->nurbs list which
gave threading issues.
Now added BKE_vfont_to_curve_nubase() which operates on a given nurbs list
base which is local in do_makeDispListCurveTypes().
By the looks of it it wouldn't give speed regression because previously it
also was creating nurbs for every font object sharing the same curve data.
Main issue here is that glBuf* calls were invoked from threads different than
main thread. This caused a crash (since those do not have a GL context active).
Fix here is twofold:
* add an ID buffer in buffer pool that handles pbvh buffers and is freed
from main thread when gpu_buffer_pool_free_unused is called.
* do not create glbuffers in derivedmesh creation routine, rather tag nodes
for update and create those in the draw function
(guaranteed to be called from main thread)
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D169
Added an API method to Text datablocks for moving the cursor to a specific line in the
file. This makes it possible to reset the cursor position at the end of the sysinfo
operator.
This creates the initial two tabs for Edit Mode to makes the new toolbar tabs usable. Further toolbar organization is to follow, but this at least makes the features work with the current organization.
This makes the panel slightly cleaner, giving more separation between the distinct elements. It also adds a new "Measure" label, so that the Ruler/Protractor is not so closely "related" to the Grease Pencil.
This adds the remaining object categories to the Create tab, including Nurb Surfaces, Lamps, Meta objects, etc. I've also updated the labeling for the various buttons to be consistent with one another.
This adds buttons in the 3D View toolbar to add new mesh objects to the scene. These buttons are added via the Create tab/category, soon to be populated with other object types as well.
A panel is added for both Object mode and Edit mode so as to not display in incorrect context (such as Pose Mode).
This updates the organization and naming of the Basic tab for the 3D View toolbar. It reorganizes the tools that were already there and adds Mirror and Duplicate Linked object operators.
Summary:
Here are the initial tabs for the 3D View Object Mode toolbar. This is based on the discussions between myself and @billrey in T37569
P.S. apologies for the duplicate. The first one I mistakenly sent to the Edit Mode task, T37568. I have corrected it and also made a couple changes, per @campbellbarton's request.
Reviewers: campbellbarton
CC: campbellbarton, brecht, billrey
Maniphest Tasks: T37569
Differential Revision: http://developer.blender.org/D129
Issue seems to be caused by thread-unsafe IMB_freeImBuf.
Now use generic BKE_image_release_ibuf to de-reference
image buffer which was acquired from image datablock.
EditFont's use of Curve.len was very confusing, in editmode it
represented the number of characters, in object mode the number of
bytes. add Curve.len_wchar and keep track of both.
Also don't convert the editmode text into utf8 on every keystroke.
Now this is done on exiting editmode or save - to match most other
object types.
This also fixes curves 'body_format' being reported with an invalid size.
Made the system around splines order a bit smarter, so
crating a segment between two splines wouldn't switch
direction if splines are selected in a way that they're
"co-linear".
It is possible to make things even smarter using active
point and so, but that i'd consider a TODO.
Made it so scopes image buffer is displayed without any additional scale
applied on them,
Further tweaks are possible here, but that i'd consider a TODO.
Mainly fixed some style warnings reported by cpplint.
Also changed how camera (un)distortion happens internally
by replacing number of channels as a template argument
with number as channels passing as function argument.
Makes code easier to follow by eliminating loads checks
how much channels are used and which argument to pass to
the template.
Clarified the tooltip/description on the LatticePoint.co property to make it
clear that it shouldn't be edited, and that LatticePoint.co_deform should be
used instead.
This mode is designed for passive 3D TV: the viewport is split
horizontally - left eye above, right eye below - but the original camera
viewport is squashed in each half (with half the vertical resolution).
This is necessary to restore the aspect ratio in the 3D output because the TV expands each half to the full screen size.
That was nothing really wrong with the old short used for
direction, but that became kinda annoying because of compiler
idiocy which considered direction might have been zero.
Using explicit dual-state flag is more clear anyway.
Summary:
Crash was happening because of fcurve modifier stack
used modifier's DNA to store temporary data.
Now made it so storage for such a thing is being
allocated locally per object update so multiple objects
which shares the same animation wouldn't run into
threading conflict anymore.
This storage might be a part of EvaluationContext,
but that'd mean passing this context all over in
object_where_is which will clutter API for now without
actual benefit for this.
Optimization notes: storage is only being allocated
if there're Cycles modifier in the stack, so there're
no extra allocations happening in all other cases.
To make code a bit less cluttered with this storage
passing all over the place added extra callbacks to
the FModifier storage which runs evaluation with the
given storage.
Reviewers: brecht, campbellbarton, aligorith
CC: plasmasolutions
Differential Revision: https://developer.blender.org/D147
File tracking.c became rather huge and annoying to
maintain and it really contains several independent
areas of motrack pipeline.
Now we've got:
* tracking.c: general-purpose functions which are used
by blender, clip editor, RNA and so.
* tracking_detect.c: feature detection functions
(blender-side, logic is still in libmv).
* tracking_plane_tracker.c: blender-side 2D tracking logic.
* tracking_plane_tracker.c: plane track tracker.
* tracking_solver.c: functions for camera solving.
* tracking_stabilize.c: 2D stabilization functions.
* tracking_util.c: utility functions for all those files
and which shouldn't be public.
Summary:
This brings up much easier termination type usage,
which for us means we might use:
ceres::Summary::IsSolutionUsable()
instead of doing manual funky enum values check.
Reviewers: keir
Differential Revision: https://developer.blender.org/D153
Summary:
Issue was caused by the same tile being written twice to
the EXR file. This was happening because of partial update
of work-in-progress tiles was merging result to the final
render result in order to make color management pipeline
happy.
We need to avoid such a merges and keep memory usage as
low as possible when Save Buffers is enabled.
Now render pipeline will allocate special display buffer
in render layer which will contain combined pass in the
display space. This keeps memory usage as low as we can
do at this moment.
There's one weak thing which is changing color management
settings during rendering would lead to lossy conversion.
This is because render result's display buffer uses color
space from the time when rendering was invoked.
This is actually what was happening in previous release
already actually so not a big issue.
Reviewers: brecht
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D162
key->elemsize is set to 16 for ID_CU (i.e. Curves and NURBS surfaces). However,
this value is only correct for NURBS (which use BPoints). When trying to keyframe
the nth vertex of a particular shape key's data (where the shape keys are being
used on Bezier curves), the RNA Paths for that are generated with the wrong
data index. From empirical testing, it appears that this should be 12 instead.
of the drawstr and replaces the middle by '...'
Rationnal:
It's usually easier to figure out what a troncated string means if you have both its start and end parts.
Details:
It currently affects all UI items (text fields, buttons, labels...).
Only exceptions are buttons inside menus, text fields & co being edited, and numbuttons/sliders.
Note that some static texts (like panels' titles or items in outliner) do not use usual UI items,
and just draw the whole text, using OpenGL to clip it. Will make another patch to fix this.
Reviewers: campbellbarton, brecht, carter2422, #user_interface
Reviewed by: brecht, carter2422
CC: billrey
Differential Revision: https://developer.blender.org/D114
is used. When numerical input was active confirmation events were not
getting handled.
Code here seems a bit duplicated, it may be possible to simplify it, but
leaving it slightly bloated for now.
This aligns the User Preferences > File > Author field and label horizontally, saving space and making it more consistent with author text fields within the same section.
This commit adjusts the spacing of the items in User Preferences > System. The spacing is adjusted via separators between each section of options. This better differentiates between groups of options. It also removes a few extra separators for more consistency.
This does not support staying fixed while the surface deforms, but for static
meshes it should match up with the surface texture coordinates. Implemented
as a matrix transform from objects space to mesh texture space.
Making this work for deforming surfaces would be quite complicated, you might
need something like harmonic coordinates as used in the mesh deform modifier,
probably will not be possible anytime soon.
The console window is hidden by default, but we need to avoid this when
starting from the command prompt, because it would hide the window you just
typed the command in.
Previously it would check if Blender was started from "explorer.exe" to
determine that, but that wasn't working for application launchers like
Appetizer or Colibri. Instead we now check if the process ID is the same as
the process ID of the console window, which appears to work reliably.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D122
Volumes can now have textured colors and density. There is a Volume Sampling
panel in the Render properties with these settings:
* Step size: distance between volume shader samples when rendering the volume.
Lower values give more accurate and detailed results but also increased render
time.
* Max steps: maximum number of steps through the volume before giving up, to
protect from extremely long render times with big objects or small step sizes.
This is much more compute intensive than homogeneous volume, so when you are not
using a texture you should enable the Homogeneous Volume option in the material
or world for faster rendering.
One important missing feature is that Generated texture coordinates are not yet
working in volumes, and they are the default coordinates for nearly all texture
nodes. So until that works you need to plug in object texture coordinates or a
world space position.
This is work by "storm", Stuart Broadfoot, Thomas Dinges and myself.
Solved by adding RW lock to BKE_vfont_to_curve.
So now all the threads are allowed to read chars from ghash,
but they'll be locked as soon as one thread would need to load
more chars from font to the ghash.
This is done using the existing Emission node and closure (we may add a volume
emission node, not clear yet if it will be needed).
Volume emission only supports indirect light sampling which means it's not very
efficient to make small or far away bright light sources. Using direct light
sampling and MIS would be tricky and probably won't be added anytime soon. Other
renderers don't support this either as far as I know, lamps and ray visibility
tricks may be used instead.
This works pretty much as you would expect, overlapping volume objects gives
a more dense volume. What did change is that world volume shaders are now
active everywhere, they are no longer excluded inside objects.
This may not be desirable and we need to think of better control over this.
In some cases you clearly want it to happen, for example if you are rendering
a fire in a foggy environment. In other cases like the inside of a house you
may not want any fog, but it doesn't seem possible in general for the renderer
to automatically determine what is inside or outside of the house.
This is implemented using a simple fixed size array of shader/object ID pairs,
limited to max 15 overlapping objects. The closures from all shaders are put
into a single closure array, exactly the same as if an add shader was used to
combine them.
This is the simplest possible volume rendering case, constant density inside
the volume and no scattering or emission. My plan is to tweak, verify and commit
more volume rendering effects one by one, doing it all at once makes it
difficult to verify correctness and track down bugs.
Documentation is here:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Materials/Volume
Currently this hooks into path tracing in 3 ways, which should get us pretty
far until we add more advanced light sampling. These 3 hooks are repeated in
the path tracing, branched path tracing and transparent shadow code:
* Determine active volume shader at start of the path
* Change active volume shader on transmission through a surface
* Light attenuation over line segments between camera, surfaces and background
This is work by "storm", Stuart Broadfoot, Thomas Dinges and myself.
* Henyey-Greenstein scattering closure implementation.
* Rename transparent to absorption node and isotropic to scatter node.
* Volume density is folded into the closure weights.
* OSL support for volume closures and nodes.
* This commit has no user visible changes, there is no volume render code yet.
This is work by "storm", Stuart Broadfoot, Thomas Dinges and myself.
Summary:
Original issue was caused by wring detection of whether
SS modifier is being applied for render/viewport. After
recent dependency graph commit this was no longer an
issue.
But it still might have happen conflict between viewport
draw and scene_update_for_newframe() invoked from render
thread when using external engines like Cycles.
Solved by adding viewport lock while scene is being
updated and objects being exported to the render engine.
Same lock was already used for Blender Internal.
Reviewers: brecht
Reviewed By: brecht
Maniphest Tasks: T36474
Differential Revision: https://developer.blender.org/D138
Issue was caused by missing objects update for temporary
freestyle objects. This happened because of the fact that
such objects doesn't have any relations, as in they're
corresponding to root nodes in the DAG.
This situation wasn't handled by DAG_threaded_update_begin()
which considered there's only one root node in the DAG.
The options are:
Base: Base mesh
Deform: shape keys and deform modifiers
Final: All deformations and modifiers
It would be nice to have a way of specifying where exactly in the
modifier stack the collision shape is generated. However this is not
staight forward since the rigid body simulation is not part of the
modifier system and would require hacks to make it work.
Summary:
Made objects update happening from multiple threads. It is a task-based
scheduling system which uses current dependency graph for spawning new
tasks. This means threading happens on object level, but the system is
flexible enough for higher granularity.
Technical details:
- Uses task scheduler which was recently committed to trunk
(that one which Brecht ported from Cycles).
- Added two utility functions to dependency graph:
* DAG_threaded_update_begin, which is called to initialize threaded
objects update. It will also schedule root DAG node to the queue,
hence starting evaluation process.
Initialization will calculate how much parents are to be evaluation
before current DAG node can be scheduled. This value is used by task
threads for faster detecting which nodes might be scheduled.
* DAG_threaded_update_handle_node_updated which is called from task
thread function when node was fully handled.
This function decreases num_pending_parents of node children and
schedules children with zero valency.
As it might have become clear, task thread receives DAG nodes and
decides which callback to call for it.
Currently only BKE_object_handle_update is called for object nodes.
In the future it'll call node->callback() from Ali's new DAG.
- This required adding some workarounds to the render pipeline.
Mainly to stop using get_object_dm() from modifiers' apply callback.
Such a call was only a workaround for dependency graph glitch when
rendering scene with, say, boolean modifiers before displaying
this scene.
Such change moves workaround from one place to another, so overall
hackentropy remains the same.
- Added paradigm of EvaluaitonContext. Currently it's more like just a
more reliable replacement for G.is_rendering which fails in some
circumstances.
Future idea of this context is to also store all the local data needed
for objects evaluation such as local time, Copy-on-Write data and so.
There're two types of EvaluationContext:
* Context used for viewport updated and owned by Main. In the future
this context might be easily moved to Window or Screen to allo
per-window/per-screen local time.
* Context used by render engines to evaluate objects for render purposes.
Render engine is an owner of this context.
This context is passed to all object update routines.
Reviewers: brecht, campbellbarton
Reviewed By: brecht
CC: lukastoenne
Differential Revision: https://developer.blender.org/D94
* UI: Remove deprecated condition (CURVE_RIBBONS) and hide backface property, when it's hardcoded in C (Curve/Line segments && Ribbons).
* Remove "use_tangent_normal" and "CURVE_KN_TANGENTGNORMAL" as its unused (follow up for last commit).
* Remove dead code from line segments BVH refine.
CURVE_KN_TANGENTGNORMAL and CURVE_KN_TRUETANGENTGNORMAL are either both true, or both false. Therefore a true/false condition is impossible.
This was a leftover of CURVE_CUSTOM, which was removed in r59234.
* Use "else if" in blender_curves.cpp.
ver=re.search(r'[0-9]+(\.[0-9]+[svn]+)+',line)orre.search(r'[0-9]+(\.[0-9]+)+',line)# read the "based on LLVM x.xsvn" version here, not the Apple version
ifenv['C_COMPILER_ID']=='gcc'andenv['CCVERSION']>='4.6.1':# for correct errorhandling with gcc >= 4.6.1 we need the gcc.dylib and gomp.dylib to link, thus distribute in app-bundle
cmd='install_name_tool -id @executable_path/lib/libgcc_s.1.dylib %s/%s.app/Contents/MacOS/lib/libgcc_s.1.dylib'%(installdir,binary)# change id of libgcc
cmd='install_name_tool -id @executable_path/lib/libgomp.1.dylib %s/%s.app/Contents/MacOS/lib/libgomp.1.dylib'%(installdir,binary)# change id of libgomp
commands.getoutput(cmd)
cmd='install_name_tool -change %s/lib/libgcc_s.1.dylib @executable_path/lib/libgcc_s.1.dylib %s/%s.app/Contents/MacOS/lib/libgomp.1.dylib'%(instname,installdir,binary)# change ref to libgcc
cmd='SetFile -d "%s)" -m "%s)"%s/%s.app'%(time.strftime("%m/%d/%Y %H:%M"),time.strftime("%m/%d/%Y %H:%M"),installdir,binary)# give the bundles actual creation/modification date
commands.getoutput(cmd)
ifenv['WITH_BF_OPENMP']:
ifenv['C_COMPILER_ID']=='gcc'andenv['CCVERSION']>='4.6.1':# for correct errorhandling with gcc >= 4.6.1 we need the gcc.dylib and gomp.dylib to link, thus distribute in app-bundle
cmd='install_name_tool -id @executable_path/../Resources/lib/libgcc_s.1.dylib %s/%s.app/Contents/Resources/lib/libgcc_s.1.dylib'%(installdir,binary)# change id of libgcc
cmd='install_name_tool -id @executable_path/../Resources/lib/libgomp.1.dylib %s/%s.app/Contents/Resources/lib/libgomp.1.dylib'%(installdir,binary)# change id of libgomp
commands.getoutput(cmd)
cmd='install_name_tool -change %s/lib/libgcc_s.1.dylib @executable_path/../Resources/lib/libgcc_s.1.dylib %s/%s.app/Contents/Resources/lib/libgomp.1.dylib'%(instname,installdir,binary)# change ref to libgcc
(BoolVariable('WITHOUT_BF_PYTHON_UNPACK','dont remove and unpack Python modules everytime if true',False)),
(BoolVariable('WITHOUT_BF_OVERWRITE_INSTALL','dont remove existing files before breating the new install directory (set to False when making packages for others)',False)),
(BoolVariable('BF_FANCY','Enable fancy output if true',True)),
is a 3D modelling and rendering package. It is the in-house software of a high quality animation studio, Blender has proven to be an extremely fast and versatile design instrument. The software has a personal touch, offering a unique approach to the world of Three Dimensions.
Use Blender to create TV commercials, to make technical visualizations, business graphics, to do some morphing, or design user interfaces. You can easy build and manage complex environments. The renderer is versatile and extremely fast. All basic animation principles (curves & keys) are well implemented.
http://www.blender.org
.SHOPTIONS
Blender 2.66
Usage: blender [args ...] [file] [args ...]
.br
.SS"Render Options:"
.TP
.B\-bor\-\-background<file>
.br
Load <file> in background (often used for UI\-less rendering)
.br
.TP
.B\-aor\-\-render\-anim
.br
Render frames from start to end (inclusive)
.br
.TP
.B\-Sor\-\-scene<name>
.br
Set the active scene <name> for rendering
.br
.TP
.B\-for\-\-render\-frame<frame>
.br
Render frame <frame> and save it.
.br
+<frame> start frame relative, \-<frame> end frame relative.
.br
.TP
.B\-sor\-\-frame\-start<frame>
.br
Set start to frame <frame> (use before the \-a argument)
.br
.TP
.B\-eor\-\-frame\-end<frame>
.br
Set end to frame <frame> (use before the \-a argument)
.br
.TP
.B\-jor\-\-frame\-jump<frames>
.br
Set number of frames to step forward after each rendered frame
.br
.TP
.B\-oor\-\-render\-output<path>
.br
Set the render path and file name.
.br
Use // at the start of the path to
.br
render relative to the blend file.
.br
The # characters are replaced by the frame number, and used to define zero padding.
.br
ani_##_test.png becomes ani_01_test.png
.br
test\-######.png becomes test\-000001.png
.br
When the filename does not contain #, The suffix #### is added to the filename
.br
The frame number will be added at the end of the filename.
is a 3D modelling and rendering package. It is the in-house software of a high quality animation studio, Blender has proven to be an extremely fast and versatile design instrument. The software has a personal touch, offering a unique approach to the world of Three Dimensions.
is a 3D modelling and rendering package. Originating as the in-house software of a high quality animation studio, Blender has proven to be an extremely fast and versatile design instrument. The software has a personal touch, offering a unique approach to the world of Three Dimensions.
Use Blender to create TV commercials, to make technical visualizations, business graphics, to do some morphing, or design user interfaces. You can easy build and manage complex environments. The renderer is versatile and extremely fast. All basic animation principles (curves & keys) are well implemented.
Use Blender to create TV commercials, to make technical visualizations, business graphics, to create content for games, or design user interfaces. You can easy build and manage complex environments. The renderer is versatile and extremely fast. All basic animation principles (curves & keys) are well implemented.
http://www.blender.org''')
@@ -119,7 +125,7 @@ while lines:
fw('''
.br
.SH SEE ALSO
.B yafaray(1)
.B luxrender(1)
.br
.SH AUTHORS
@@ -128,4 +134,5 @@ This manpage was written for a Debian GNU/Linux system by Daniel Mester
<cyril.brulebois@enst-bretagne.fr> and Dan Eicher <dan@trollwerks.org>.
@@ -34,7 +34,7 @@ base class --- :class:`PyObjectPlus`
:type:int
..attribute:: walkDirection
The speed and direction the character is traveling in using world coordinates. This should be used instead of applyMovement() to properly move the character.
@@ -49,7 +49,7 @@ A quick list of helpful things to know before starting:
* Blender uses Python 3.x; some 3rd party extensions are not available yet.
* The interactive console is great for testing one-liners, It also has autocompleation so you can inspect the api quickly.
* The interactive console is great for testing one-liners, It also has autocompletion so you can inspect the api quickly.
* Button tool tips show Python attributes and operator names.
@@ -247,7 +247,8 @@ Examples:
Operator Poll()
^^^^^^^^^^^^^^^
Many operators have a "poll" function which may check that the mouse is a valid area or that the object is in the correct mode (Edit Mode, Weight Paint etc). When an operator's poll function fails within python, an exception is raised.
Many operators have a "poll" function which may check that the mouse is in a valid area or that the object is in the correct mode (Edit Mode, Weight Paint etc).
When an operator's poll function fails within python, an exception is raised.
For example, calling bpy.ops.view3d.render_border() from the console raises the following error:
``code.interact`` can be added at any line in the script and will pause the script an launch an interactive interpreter in the terminal, when you're done you can quit the interpreter and the script will continue execution.
#warning "Eigen2 support is deprecated in Eigen 3.2.x and it will be removed in Eigen 3.3. (Define EIGEN_NO_EIGEN2_DEPRECATED_WARNING to disable this warning)"
#else
#pragma message ("Eigen2 support is deprecated in Eigen 3.2.x and it will be removed in Eigen 3.3. (Define EIGEN_NO_EIGEN2_DEPRECATED_WARNING to disable this warning)")
#endif
#endif // EIGEN_NO_EIGEN2_DEPRECATED_WARNING
#include "src/Core/util/DisableStupidWarnings.h"
/** \ingroup Support_modules
* \defgroup Eigen2Support_Module Eigen2 support module
* This module provides a couple of deprecated functions improving the compatibility with Eigen2.
*
* \warning Eigen2 support is deprecated in Eigen 3.2.x and it will be removed in Eigen 3.3.
*
* This module provides a couple of deprecated functions improving the compatibility with Eigen2.
*
* To use it, define EIGEN2_SUPPORT before including any Eigen header
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
