There were two issues. First, I made a mistake when I switched from unsigned
to signed integers. Second, two classes with the same name were defined in
separate files. Those classes are in an anonymus namespace now, so that they
don't leak into other files.
ad4928a171 disabled alignment for too many cases. Still try to avoid
aligning many items, to avoid thousands of redundant alignment
calculations. But now we're much more picky adding an sub-row with
alignment.
In Blender 2.90 EEVEE materials were refactored that introduced crashes on Intel
GPUs on Windows. The crash happened in the `local_context_workaround` that temporary
stored compiled materials in a binary form to reload it in the main GL context.
It has been tested that the workaround isn't needed anymore for HD6xx GPUs, but it
is still needed for HD4000.
After several unsuccesfull fixes we came to the conclusion that we could not support
the local context workaround and needed to come with a different workaround. The idea
of this patch is that in these cases there is only a single context that is used for
rendering. Threads that uses these contextes are guarded by a mutex and will block.
Impact on User Level:
* Due to main mutex lock the UI freezes when rendering or baking or feel less snappy
Reviewed By: Clément Foucault, Brecht van Lommel
Differential Revision: https://developer.blender.org/D8410
As far as I can see, this problem was introduced with the gpu_free_unused_buffers() changes in rB97b597c.
The code checks the returned pointer to stop looping, but the last iteration will return the last GPUTexture and set the queue to NULL, meaning that the next pop() will crash.
Differential Revision: https://developer.blender.org/D8468
Some implementation have different maximum texture size.
This patch avoid crash when texture allocation fails when:
- trying to bake a lightcache too big for the OpenGL imeplementaion.
- loading a cache from file that is too big for the OpenGL imeplementation.
Face Sets where only set and updated on the PBVH after starting a sculpt
tool. In order to preserve the visibility they store when changing
levels, they need to be updated and sync also on PBVH creation
Reviewed By: sergey
Maniphest Tasks: T78665
Differential Revision: https://developer.blender.org/D8225
Duplicates of a grid corner adjacent to an edge which are on the
adjacent grid of the same face were not added when requested.
Needed for D8356 to work, it may also fix some other bug in Multires.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D8385
The cloth brush has a defined simulated area with a falloff. In the falloff
area (the area between the dashed white circle and the exterior white
circle), simulation properties change in order to fade out the
simulation deformation effects towards the boundary.
With some brushes and stroke types (like anchored strokes with pinching
or grabbing with full strength), it is possible to apply more force than
what the boundary falloff can compensate, so the simulation breaks when
this happens.
This option pins the falloff area with softbody constraints, This
produces a much better deformation falloff and it is no longer possible
to move the vertices near the simulation boundary, so the simulation
won't break no matter the strength of the forces applied inside the
simulated areas.
This is an option as it is particularly useful for some brushes to add
localized details, but for brushes that are supposed to deform the
entire mesh (like the grab brush in D8424), this can add unwanted
softbody constraints that affect the simulation result.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D8435
Box mask is not a selection, so it should not be part of the select
operator. This allows to add more sculpt mode specific functionality and
properties and to share more code with the lasso mask operator in a
later refactor.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D8456
When switching workspaces we need to have an unused screen layout that
we can activate. The other window now showed the only available screen
layout in fullscreen though.
Usually when there's no unused screen layout we duplicate an existing
one, but that code didn't respect the fullscreen case properly.
This also tries to clean up the logic a bit, but things are still rather
complicated to follow.
Changes in this code are always risky. Of course things worked fine in
my tests, but I wouldn't be surprised if something breaks.
Filtering Collections when also filtering object children would only
display the parent object. Add a check for the NO_CHILDREN filter before
creating the object-parent hierarchy.
The winding order of the faces changes when flipping the faces.
This lead to the loop indices changing as well.
Now we take this into account when restoring and flipping the custom
normals. Before the normals would be swapped.
The winding order of the faces changes when flipping the faces.
This lead to the loop indices changing as well.
Now we take this into account when restoring and flipping the custom
normals. Before the normals would be swapped.
The problem in this case was that the flag for active fire was not set. With hidden flow sources the flag was not updated in update_flowsflags().
The solution for this is to take the active field from the config cache file.
Previously a for loop with two iterations was used to calculate the
3D segment locations for the input number and the higher power of 2.
Splitting off the inside of the for loop to a separate function makes
the code more readable.
This commit also includes a simple timer for bevel, enabled with a
define. Interestingly, the cleanup in this commit happended to give
a 3% speedup on a Ryzen 3700x for a bevel calculation with 64
segments.
Same exact system as instancing on vertices for a mesh. More powerful ways
of instancing will be added at some later point, but this makes the basics
works and is consistent with other geometry types.
