Blender Git Loki

Cycles: Optimization for Heterogeneous Volume Shadows.

* Don't compute expf() for every step, instead sum the intermediate values and calculate it every N (8 for now) steps.

This helps a few percent (~5% on a cube with wave texture) in my tests here.

Multithreading for hair simulation.

The hair solver now uses the generic task BLI_taskpool implementation
to enable threaded stepping of the simulation. For now the hair data
can very easily be split into tasks, since all hairs are entirely
independent. Later on this becomes more restricted through hair-hair
collisions, but there are some advanced techniques for dealing with it
(static pair pruning etc.)

OpenSubdiv: Fix crash entering sculpt mode

Fixes:

* Space sticky did not call search menu
* Cancelling and confirming with no active button would respawn the pie
menu immediately in drag style interaction
* Area now covers the whole screen, collision with pie buttons is done
always.

Merge branch 'master' into opensubdiv-modifier

OpenSubdiv: Support GPU tessellation in edit mode

This should speed up deformation-only modifications in edit
mode very well (assuming subsurf is the last modifier in the
stack).

This is so-called "seems to work" commit, requires much deeper
tests, mainly shading/texturing-wise.

single-threaded re-runs work ok

Fix issue with navigation pie/search menu

Removed some leftover debug print code.

Set a flag in the hair modifier whenever the hair data in the solver
becomes invalid.

This is needed when the number of hairs or points changes, but also
in case on sudden positional changes etc. (like resetting to rest
position), because the solver would otherwise calculate a large delta
and generate a lot of fake energy.

back to single-threaded method to debug manta re-run errors

Change the enum fetching code to be order independent.

This is done by checking each item in a dynamic enum for existence.

This can result in n^2 cost in worst case (for 8 items about 64 steps)
but since this is done once on initialization and makes the system more
robust, better prefer it.

Fix problem of hair root animation by keeping the solver around as a
runtime instance.

The SolverData also functions as a short-term history for constructing
the hair root animation/deformation curves. This is necessary because
the DerivedMesh data on its own does not have interpolation intervals
nor velocity information.

It also means that we have to keep this data alive over multiple frames
in order to do root animation properly. So now the solver is created as
a runtime instance.

Code cleanup, move function to correct header

Base support for hair root animation.

Note: This does not work currently, because of the way the solver is
reconstructed in every time step currently, and so does not provide
actual changes in the transform.

Note2: The mesh sampling system does not support changes in DerivedMesh
topology so far. It can be necessary to move the hair system to the
bottom of the modifier stack, because the hair roots are created for
the final mesh version and for animation support it can only use the
mesh being passed to the modifier. Any later subsurf etc. (constructive
modifier) will alter the topology and cause a difference in the DM used
for hair roots vs. DM used for calculating deformation.

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

Quiet double promotion warnings

Merge branch 'master' into pie-menus

Change rename to one proposed by Campbell

Address Dalai's concerns