Blender Git Commits

Hair System test

Fixup for stereo output (side-by-side, top-bottom)

Multiplying by inverse obmat for main particles is redundant, since
the hairmat already places them in object space.

Add simple code to add (ctrl-P) and display parents. (WIP)

This is not a serious attempt though and possibly a good solution should
go through a dependency graph. Also there are some crashes on undo.

Partial revert of the particle copy operator: use base data again
instead of the interpolated curves.

Otherwise usually far too many points are created for the amount of
detail required.

Copy a number of settings from old particle hair to the new system,
where they are compatible.

Make the "copy from particles" operator for hair always clear the system
first, so replacing hair becomes easier.

Implemented materials for hair systems.

This uses basically the same system as current particle hair: the hair
system is linked to the material via a slot index, which refers
to the object material slots. Unlike particle settings the hair system
is (for now) not a separate datablock, so the mapping is a bit more
consistent.

Implemented geometric hair rendering in Cycles ("Triangles" + "Thick"
method).

Unlike the particle hair this uses the existing smoothed frame provided
by the hair system. This gives better results in corner cases, where
twisting of the strand would be an issue otherwise due to undefined
bending axes, and can also be calculated quite cheaply using linear
frame interpolation.

Implemented "Ribbons" method for Cycles hair rendering.

Fixed compilation under MSVC

added comments

removed commented parts

inflow without noise; passing rotated geometry to amantaflow correct

Added select-linked, fixed leak, cleaned up operator poll functions

Enable "generated" coordinates for hair curves in Cycles.

Also includes a fix for attributes: the buffer retrieved for attributes
is not automatically resized when hair curves try to add the same
attribute as particles at a later time ... This should have more robust
behavior!

Stereo Output for Multilayer when rendering or saving from Image Editor

We now do stereo-magic to ALL layers of multilayers.

The only remaining part of stereo output is for videos when saved with
individual views (the naming scheme at the moment is handled inside the
movie code), to be tackled later (TODO).

Stereo output core - support rect/rectf instead of ImBufs only (for openexr multilayer)

Fix sanitized memory for rna_camera

(thanks to -fsanitize=address and Lukas Toenne ;)

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