Blender Git Commit Log

Cycles: Fixed Cryptomatte crash when some RenderLayers use Cryptomattes and some don't

The kernel level pass_aov array was not re-initialised properly and kept the info from the previous RenderLayer instead

Smooth new points while drawing

A new smooth algorithm is used for each new buffer point using the last two points to calculate the estimated position and average this point with the current mouse position.

This smooth reduce the jitter of the stroke and allows to use lower post smooth values to reduce stroke change after finish stroke.

Now the factor is hardcoded, but after some more testing maybe we need add a parameter by brush.

Merge branch 'master' into id_override_static

Conflicts:
source/blender/editors/interface/interface_handlers.c

Fix T53250: Crash when linking/appending a scene to a blend when another linked scene in this blend is currently open/active.

Inner DAG code would not check against NULL pointer, and in case of an
active linked scene, scene pointer will be NULL here, so we have to
check it ourself. ;)

Depsgraph: Cleanup, remove obsolete unused function

epsgraph: Use explicit graph API for sequencer

Depsgraph: Use explicit graph API for point cache

Depsgraph: Use explicit graph API for audio bake

Depsgraph: Use explicit graph API for OpenGL render

Depsgraph: Use new API for alembic C-API

Update comment, reference new API function

Depsgraph: Use new scene update API in preview render

Depsgraph: Make graph to be per-scene-layer for RNA

New access is C.scene.render_layers.active.depsgraph. This will give depsgraph
for a given layer. In the future there will need to be some extra context to be
passed.

Depsgraph: Cleanup, remove depgraph_legacy from evaluation routines

Depsgraph: Cleanup, remove depsgraph_legacy from building routines

Depsgraph: Cleanup, remove unused function

Depsgraph: Cleanup, remove depsgraph_legacy from depsgraph_tag.cc

Depsgraph: Avoid non-requested allocation of depsgraph

Cycles: reduce closure memory usage for emission/shadow shader data.

With a Titan Xp, reduces path trace local memory from 1092MB to 840MB.
Benchmark performance was within 1% with both RX 480 and Titan Xp.

Original patch was implemented by Sergey.

Differential Revision: https://developer.blender.org/D2249

Code refactor: sum transparent and absorption weights outside closures.