Blender Git Loki

Cycles: Implement texture size limit simplify option

Main intention is to give some quick way to control scene's memory
usage by clamping textures which are too big. This is really handy
on the early production stages when you first create really nice
looking hi-res textures and only when it all works and approved
start investing time on optimizing your scene.

This is a new option in Scene Simplify panel and it acts as
following: when texture size is bigger than the given value it'll
be scaled down by half for until it fits into given limit.

There are various possible improvements, such as:

- Use threaded scaling using our own task manager.

This is actually one of the main reasons why image resize is
manually-implemented instead of using OIIO's resize. Other
reason here is that API seems limited to construct 3D texture
description easily.

- Vectorization of uchar4/float4/half4 textures.

- Use something smarter than box filter.

Was playing with some other filters, but not sure they are
really better: they kind of causes more fuzzy edges.

Even with such a TODOs in the code the option is already quite
useful.

Reviewers: brecht

Reviewed By: brecht

Subscribers: jtheninja, Blendify, gregzaal, venomgfx

Differential Revision: https://developer.blender.org/D2362

GPU: Consider latest Gallium driver an official ATI/AMD

This will make triple buffer used by default for such configuration.

Ideally we would switch to triple buffer on all platforms, but let's
do it in 2.8 branch and don't open can of worms in master now.

This should solve issues like T49945.

Merge remote-tracking branch 'origin/master' into soc-2016-cycles_denoising

Conflicts:
intern/cycles/blender/blender_session.cpp

Cycles: Enable Flush-to-zero for denormals on Windows

Cycles: Fix Denoising Pass detection when loading Multilayer EXRs

Cycles Denoising: Add standalone animation denoising call to the Cycles Python API

Previously, a build of Cycles Standalone was needed for animation denoising.
Now, it is possible to call it from Python Scripts in Blender or directly from the command line (through --python-expr).

Cycles Denoiser: Store debug info in one exr per tile instead of many pfms

Cycles Denoising: Fix pass handling with MSVC

Cycles: Move code to new and nicer file layout

Cycles: Remove old WLR filter

Cycles: Get rid of one loop over all denoised pixels by using backsubstitution instead of explicit inversion

Cycles: Move the final_pass functions to a unified implementation and generally shuffle a lot of code around

Cycles: Improve robustness of the Cholesky decomposition

Cycles: Implement Cross-Denoising

Right now this is pretty bad, but it can be made much better.

Glog: Fix compilation error on ppc64el

This was fixed in upstream already. Time to re-bundle?

CLEW: Workaround compilation error on ppc64el

Something was conflicting here, causing C++ to consider bool as
a __vector(4) bool.

Partly revert own rBb97c567c1df1e, clear_proxy is actually safe.

This is very confusing, in fact, and rna tooltip was wrong,
BKE_object_make_local_ex actually ensures we never have several proxies
of same object, since it always clears proxy when it has to copy object
to make it local...

What that RNA function is probably missing, though, is same logic as in
BKE_library_make_local to actually remap proxy from old linked object to
new local one.

Layers - initial commit

This is mostly DNA + doversion and some util functions.
I need to look at the layer-manager branch to see how
Julian Eisel (Severin) managed to handle nested lists
(required for collections).

Also, since I think doversioning may get more complex as the project
evolves, I started the vesioning_280.c file.

And the changes in readfile.c were so that I could include into the
versioning_280.c file only the DNA_* that were really needed.

(if someone feels like doing a cleanup, I'm sure the other
versioning_*.c files could use the same treatment)

Fix two very bad issues in new ID.make_local RNA function.

I) `clear_proxy` parameter was not assigned to parm in RNA define code,
so 'pyfunc optional' flag was set to `new_id` parameter of `user_remap`
func - super ugly!

II) `clear_proxy` parameter itself, when set to False, would allow to
leave .blend file in invalid state (more than one proxy of same object),
this should never, ever be allowed in RNA API imho. Left the PAI
untouched for now, just disabled any effect from this parameter (hence
always clearing proxy when copying).

Cleanup: get rid of unused `BKE_constraints_relink()`.

Libquery/generic ID remapping code handles this now.