Blender Git Commits

code cleanup: grease pencil eraser had duplicated logic for getting screen coords of a point, move into a static function.

code cleanup: use checks for empty rather then size in the BGE

fix for logical errors
- range check on hair_velocity_smoothing() was off by one.
- cloth sim parm's are used before NULL check in readfile.c

style cleanup: if();

code cleanup: glare stream operation was setting alpha array twice.

add type checking for more error prone macros.

rename avi files (endian.h is common linux include).

patch [#32556] Stupid endian conversion in avi format
from Andreas Schwab (schwab)

modified to use code from BLI_endian_switch.

Optimization for endian switching, but shifting is a lot faster then using a temp char (approx 18x speedup on my system).

patch [#32791] Spelling mistakes corrected.
from Sunny Gogoi (darkowlzz)

Fix render engine API compatibility breakage in end_result, this parameter
should have been optional.

BGE: Committing patch [#31442] "API improvements: Group references python api, for better control over groups and instances" by Martin Sell (moerdn). This patch adds a member and a group property to KX_GameObject:
* KX_GameObject.member returns the list of group members if the object is a group object, otherwise None is returned
* KX_GameObject.group returns the group object that the object belongs to or None if the object is not part of a group

Merged changes in the trunk up to revision 51126.

Conflicts resolved:
source/blender/blenloader/intern/readfile.c
source/blender/windowmanager/WM_types.h

Force USE_QTKIT for osx 64bit

Fix for WITH_LLVM cmake option: When using WITH_CYCLES_OSL this option is forced ON, but this change was not cached yet. Thanks to Jens Verwiebe for providing this fix.

Avoid unecessary minimizer calculations in dualcon

* The minimize() function, which solves a least-squares problem, is
only needed for sharp remesh mode, but was being calculated for
smooth and blocks modes as well. Disabling this calculation when
it's not needed gives a big performance boost.

Code cleanups for dualcon octree

* Move InternalNode operators from Octree class into InternalNode
struct

* Constify various member functions

Increase maximum octree depth to 12

Note that this is just an RNA change, underlying dualcon octree
already supports even higher values.

Cycles / OSL:
* OSL UI message did not show up when device type was GPU, but User Preferences were None. Also remove experimental check, more convenient for testing.

Improve flat-shaded VBO drawing for sculpt meshes

Separate vertex copies are now made for flat-shading, such that the
normal is correctly flat-shaded. The element index buffer is not
created in this case.