Blender Git Commit Log

LineArt: Modifier UI cleanup.

GPencil: Change maximum limit of Join stroke

Also, clarify comments.

Merge branch 'master' into greasepencil-object

Merge branch 'master' into greasepencil-edit-curve

Merge branch 'blender-v2.91-release'

Fix T82265: GPencil interpolate crash when next frame was NULL

If the next frame of the layer was NULL and it was not the active layer crashed.

LineArt: UI logic fixe for object line art panel.

LineArt: Main panel UI optimizations.

LineArt: COLLECTION_HAS_OBJECT_CACHE_INSTANCED flag clear

Merge remote-tracking branch 'origin/master' into lanpr-under-gp

Merge branch 'geometry-nodes' into geometry-nodes-attribute-nodes

Geometry Nodes: Rename node category based on functionality

Naming based on supported geometry components only works in some cases.
For nodes that work with multiple data types, it's better to categorize
them based on their functionality.

Merge branch 'master' into geometry-nodes

Geometry Nodes: Add initial empty nodes for basic attribute workflow

This adds the boilerplate code for three nodes: "Create Attribute,"
"Random Attribute," and "Attribute Math." Combined, they should
be enough for the process of creating randomized "Rotation" and
"Scale" attributes on point clouds for instancing done later. The
implementation details are not resolved at this point, but this
starting point should make sense.

Tried to fix sculpt vcol paint undo. It's better, but I still need to
go through and thorougly analyze just what the undo stack is doing.

Paint brush no long SCULPT_undo_push_node per dab (except for first
stroke) for dyntopo. Instead it updates the original vertex color
customdata layer.

Calling into the undo system destroys threading with dyntopo, as its
undo code is single-threaded.

Try to make SCULPT_orig_vert_data_init avoid allocating undo nodes,
which calls into BMLog which, while now threadsafe (ish?) causes
threads to get bogged down in lock contention.

Cycles: Implement tile stealing to improve CPU+GPU rendering performance

While Cycles already supports using both CPU and GPU at the same time, there
currently is a large problem with it: Since the CPU grabs one tile per thread,
at the end of the render the GPU runs out of new work but the CPU still needs
quite some time to finish its current times.

Having smaller tiles helps somewhat, but especially OpenCL rendering tends to
lose performance with smaller tiles.

Therefore, this commit adds support for tile stealing: When a GPU device runs
out of new tiles, it can signal the CPU to release one of its tiles.
This way, at the end of the render, the GPU quickly finishes the remaining
tiles instead of having to wait for the CPU.

Thanks to AMD for sponsoring this work!

Differential Revision: https://developer.blender.org/D9324

Cleanup: clang-format

Previous check seems to have a race condition, resulting in Wintab not being enabled.