Blender Git Commits

Little tweaks for ifdefs.

Fixes for Fermi.

Use some more cuda_pop_context(); calls to make following this easier.

Free the CUtexObjects properly.

Only have one bindless_mapping vector, and also load textures for shader() kernel.

This fixes float4 textures o/

Updated OpenHMD to latest git version
Now uses threading for updating the hmd, better polling rates for performance on heavy scenes/low framerates

Revert "Decklink test build"

This reverts commit c769b921a19c42dbb14056e019ba61e345a791bb.

Decklink test build

Add BF to contributors list

Let's not ask for legal trouble ;)

Copy BKE_facemap stuff from wiggly-widgets branch

Idea is to split face maps of from wiggly-widgets into this branch to prepare review/merge.
Used term "facegroup" for file names since I plan to rename "face maps" to "face groups" anyway.

Revert "Decklink test build"

This reverts commit 5ef966e3978eb1abccd5dc32790d9fe119394b61.

Decklink test build

Merge remote-tracking branch 'origin/master' into experimental-build

Merge branch 'master' into wiggly-widgets

Conflicts:
source/blender/blenloader/intern/versioning_270.c

Register widget group types more like operator types

Is more consistent and I think it makes it easier to understand the role of widget groups.

Adds file space_node/node_widgets.c

Fix again assembler version of atomic_add_uint32 and atomic_sub_uint32

The assembler version returns the previous value of the variable while
all the other versions return the new value.
This was fixed before but somehow it came back when the atomic module
was refactored.

Merge remote-tracking branch 'origin/master' into decklink

Conflicts:
intern/atomic/atomic_ops.h

VideoDeclink: change default cache size to 4 frames.

Previously, cache size was only 3 frames, which caused problems with older
types of Decklink cards (was working with the 4K Extreme).
The default is now 4 frames and it is possible to specify a different cache
size when opening the card:
VideoDecklink("<mode>/<pixel>[/3D][:<cachesize>]", cardIdx)

Ex with 8 frames in cache:

VideoDecklink("pal/2vuy:8", 0)

Bindless Textures: Move bindless_mapping to CUDADevice.

don't forget to sample particles on first frame