Blender Git Commits

Merge branch 'master' into soc-2013-depsgraph_mt

Conflicts:
source/blender/blenkernel/intern/constraint.c
source/blender/blenkernel/intern/depsgraph.c
source/blender/blenkernel/intern/object.c
source/blender/blenkernel/intern/scene.c
source/blender/blenkernel/intern/shrinkwrap.c
source/blender/collada/AnimationExporter.cpp
source/blender/editors/space_image/image_ops.c
source/blender/editors/space_view3d/view3d_view.c
source/blender/makesrna/intern/rna_scene.c
source/blender/modifiers/intern/MOD_util.c
source/blender/render/intern/source/pipeline.c

Cycles: remove disabled kernel_volume_get_final_homogeneous_extinction code.

I'm not quite sure why this would be needed. It gives per color channel sigma,
which we should have so we can compute per channel attentuation without noise
instead of the single float now.

However I think this should be handled at a deeper level by making sigma a
float3 everywhere instead of doing some kind of correction for this afterwards.

Cycles Volume: add appropriate shader contexts for OSL.

This is needed to avoid using same memory block twice when the closures from
a shader evaluation are in use at the same time as the closures from another
evaluation.

Cycles Volume: remove some unnecessary rngs from shadow computation.

* New Volume Absorption node was missing in Node Editor menu.

Cycles Volume: implement volume absorption node.

This is the transparent volume node renamed, currently it is basically the
same as a scatter volume node with anisotropy 1, so it scatters light perfectly
forward.

This also does some tweaks to the henyey-greenstein closure code to avoid
division by zero and to make eval/sample consistent in some corner cases, and
some other code cleanup related to volume shaders.

Merge branch 'master' into soc-2013-dingto

Merge branch 'master' into soc-2013-depsgraph_mt

Fix compilation error when using scons

Rename evaluation_context to eval_ctx

Only affects on variable name, structure name stays the same.
Addresses code review from Campbell to make variable names
which are being passed all over be short.

Not entirely happy with this, but well ok. It's still verbose
enough and naming kind of consistent to what Python uses.

Discussed with Brecht and he's also fine with such a change.

Hopefully this is last change to be done before merge :)

Merge branch 'master' into soc-2013-depsgraph_mt

Conflicts:
source/blender/editors/render/render_internal.c
source/blender/modifiers/intern/MOD_smoke.c

Address notes from Campbell's review

- Cleanup naming in some places.
- Get rid of _ex functions with use_threads argument
in scene update routines.

Using Py_BEGIN_ALLOW_THREADS from frame_set and
update_tagged functions seems made the trick.

- Fix for possible uninitialized members of
evaluation context when it's allocated in stack.

Merge branch 'master' into soc-2013-depsgraph_mt

Address review from Brecht

* Removed residual of derivedRender code from convertblender.c
* Added check for oldcreen->scene != screen->scene before rebuilding
the DAG when changing the scene.
* Made blender compilable with smoke disabled.
* Typo fixes.

Merge branch 'master' into soc-2013-depsgraph_mt

Conflicts:
release/scripts/startup/bl_ui/properties_data_curve.py

Fix typo which was already fixed in trunk actually

Code cleanup to address Lukas codereview comments

Merge branch 'master' into soc-2013-sketch_mesh

Conflicts:
release/scripts/addons
source/blender/blenloader/intern/readfile.c
source/blender/blenloader/intern/writefile.c
source/blender/editors/object/object_modifier.c
source/blender/makesrna/intern/rna_modifier.c
source/blender/modifiers/intern/MOD_laplaciandeform.c

Workaround for render settings with mballs and cycles

Not ideal solution, but it'll work just fine for until
CoW is implemented (which is gonna to take a while anyway).

Pass evaluation context to BKE_scene_base_iter_next

This is needed to make dupilist being created with
proper render settings when counting mballs.

Without this change enabling particles for rendering
and disabling them for viewport would lead to situation
when mball used as dupliobject for this particle system
is never visible.

Also removed hardcoded evaluation context from some
RNA callbacks.

TODO: Cycles will use viewport render settings for
mballs, don't ask why. It is to be solved.