Blender Git Commit Log

macOS: Port to arm64.

Untested and not fully featured.
Missing are:
* several Python modules, notably ctypes and numpy
* Embree
* OpenImageDeniose
* Audaspace (depends on numpy)

To build it, configure with the following flags:
OSX_ARCHITECTURES=arm64
CMAKE_OSX_ARCHITECTURES=arm64
WITH_CYCLES_EMBREE=OFF
WITH_CPU_SSE=OFF
WITH_AUDASPACE=OFF

Removed extra vertices in the new bmesh

Also considering to use `BM_mesh_bm_to_me_for_eval` to avoid
time spent in `BM_mesh_bm_to_me` for mesh data not required for OBJ.
So some commented code is present.

UI: UV: Add Select Shortest Path to menu

This matches the 3D Viewport

Fix T78662: Cycles baking fails if denoising is enabled, after recent changes

This is not supported yet.

Fix build error building without OpenImageDenoise

Commit a basic working importer, with some extra-to-be-removed code.

Merge branch 'master' into soc-2020-io-performance

Merge branch 'master' into property-search-ui

Fix T78801: Eevee missing setting to enable/disable freestyle per view layer

This was only visible when Cycles was enabled.

UI: Improve transform constraint layout

- Remove the "mapping" subpanel and moves the source axis
selection ot the destination subpanel.
- Rename "Source" and "Destination" to "Map From" and "Map To" to
make the action more clear
- Gray out source axes when their data isn't selected.

These changes were discussed in D8041.

Revert "Fix T78296: Performance - Use Binary Search for MDeformWeight"

This reverts commit 39b525e0f07fa25dcda54226ade789959b642dec and
3121015dceb1d269d79690c8f15c8e1406c9b09f as tests are failing.

GPencil: Fix unreported Use Falloff curve for active frame not working

For the active frame it was using always a value of 1.0 and it was not using the curve.

Generator panel improvements, fix memory leak

Merge branch 'master' into lanpr-under-gp

Merge branch 'master' into greasepencil-edit-curve

Conflicts:
source/blender/blenloader/intern/versioning_290.c

Cycles: support OpenImageDenoise in final renders

Performance is not great currently due to the API not seeming to support
efficient denoising of multiple tiles at the same time. So in many cases
only one or a few threads will actually be denoising at the same time.

In renders with many samples this is not a big problem, but for faster
renders it's a signficant overhead.

We should try to optimize this still, possibly by batching denoising of
a bigger neighborhood of multiple tiles at once.

Cleanup: reduce hardcoded numbers in denoising neighbor tiles code

Cleanup: minor refactoring in Cycles update detection code

Cycles: optimize camera inside volume tests

Only run when there are volumes in the scene, and compute in parallel.

Ref T56939

Differential Revision: https://developer.blender.org/D8261

Fix T78776: Cycles OpenCL error after recent changes for holdouts