Blender Git Loki

Cycles: Overload DeviceTask::update_progress to take multiple tiles

`update_progress` will only update the tile sample once every second,
which is a problem when there are multiple tiles and calls to this
function. The overload allows for multiple tiles to be updated at once
by getting around the rate limit with a single call.

Cycles: Add SampleRange struct

This struct will be used to describe the ranges of tile samples passed
to the kernel when rendering multiple tiles samples at once.

Cycles: Have CPU devices use new the acquire_tiles function

This is mostly for testing at the moment, as the default work request
sizes should result in only one tile being acquired at once. May need to
tweak the request sizes later, but for now this should do.

Cycles: Add functions to acquire multiple tiles at once

Devices will render multiple smallish tiles (and eventually multiple
samples) at once to keep busy. Using smaller tiles has a few benefits:
there's less of an issue of tiles being cropped at image edges, multiple
devices can share work better, and we may even be able to increase
coherency from rendering multiple samples.

Apply collision responce on state before pre-collision solve (breaks selfcol)

This uses the pre-collision solve result only to find the collisions and
calculate the responce impulses, but rolls back to before the
pre-collision solve when it is time to actually apply the responce.

This prevents the cloth from undergoing a double solve per time-step,
which essentially made colliding clothes move much faster than
non-colliding clothes.

Cycles: Implement new NLM kernels for CUDA

Remove Bases onces and for all

This officially makes the viewport not draw anything, until we get the
new ObjectBase * to use.

This is easily revertable, but for now I prefer to make sure this is not
in the way of refactoring.

Remove weird impulse scaling

Updated python files to use select_get/_set() [not addons]

Merge branch 'master' into id_override_static

Conflicts:
source/blender/makesrna/RNA_types.h

Merge branch 'master' into asset-engine

Conflicts:
source/blender/blenloader/intern/readfile.c
source/blender/makesdna/DNA_ID.h

Fix crash.

Fix collision extra solve update and remove redundant result applies

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

Cycles Denoising: Use better and faster NLM implementation for feature pass prefiltering

Previously, the prefiltering NLM kernel was implemented just as it's described in the paper:
For every pixel P, loop over every pixel Q in the search window. Then, loop over the small patches around them, calculate the average difference, and use that to compute the weight of Q for the denoised result at P.

However, that gives you a time complexity of O(N^2 * R^2 * F^2), where N is the image size, R the search window and F the patch size...
So, this patch implements the clever idea from "A Simple Trick to Speed Up and Improve the Non-Local Means" - by reformulating the loop, it's actually possible to skip a lot of computation and replace it with a separable box filter convolution. This reduces complexity to O(N^2 * R^2 * F), and the amount of pixel differences calculated even to O(N^2 * R^2)!

Furthermore, by applying a second box-filter pass after calculating the weights, we get the "patchwise NLM" improvement basically for free!

This is CPU-only so far, but that will change soon.

Cycles: Clean up passing of int4 to CPU kernels

Since the type of int4 depends on whether SSE is enabled, the SSE kernels expect a different type than the device code.
Therefore, the content must be passed as a pointer...

Cycles Denoising: Remove now-useless feature utilities

Separate SSAO code & add orthographic support

Added saturation and value slider + fix versioning

Split interp_weights_face_v3 into specific functions for tris and quads