Blender Git Loki

Fix T52871: beauty fill error

Only lock tri's facing different directions.
Needed because scanfill creates zero area faces.

Beauty fill was skipping small faces

Correct gtest error in recent beautify change

Polyfill Beautify: half-edge optimization

Was using an edge hash for triangle -> edge lookups,
updating triangle indices for each edge-rotation.

Replace this with half-edge which can rotate edges much more simply,
writing triangles back once the solution has been calculated.

Gives ~33% speedup in own tests.

GTest: initial BLI_heap test

Use BLI_heap_reinsert for decimate and beautify

Improves performance for high poly meshes,
~70% faster for decimate, only ~10% for beautify.

BLI_hash: add BLI_heap_reinsert

Allows avoiding remove/insert calls.

BLI_heap: minor changes to the API

Recent addition of 'reinsert' didn't match logic for ghash API.

Rename to BLI_heap_node_value_update,
also add BLI_heap_insert_or_update since it's a common operation.

Missing from last commit

Fix T53632: Objects outside scene shown in popup

Only show objects in current scene when not pinned.

This commit adds a filter argument to id-template
since we may want to filter by other criteria.

Merge branch 'blender2.8' into hair_guides

Fix T53678: Smart Project UV margin ignores units

Fix an error on last commit

`count` here is multiplied by 4.

Fix crash when you change the curve resolution in Edit Mode of the Blender Render engine

Same problem as previous commit, but now in `drawobject.c`.
Also added a comment to the equivalent line in `draw_cache_imp_curve.c`.

Curve Edit Mode: Fix crash when the U-resolution of the curve is changed

`normal_len` did not follow the value of the loop that is executed to add the vertices, being different from the vbo size

Merge branch 'blender2.8' into greasepencil-object

Fix: MSVC2013 compile problems with the "inline" keyword

Cycles: CUDA support for rendering scenes that don't fit on GPU.

In that case it can now fall back to CPU memory, at the cost of reduced
performance. For scenes that fit in GPU memory, this commit should not
cause any noticeable slowdowns.

We don't use all physical system RAM, since that can cause OS instability.
We leave at least half of system RAM or 4GB to other software, whichever
is smaller.

For image textures in host memory, performance was maybe 20-30% slower
in our tests (although this is highly hardware and scene dependent). Once
other type of data doesn't fit on the GPU, performance can be e.g. 10x
slower, and at that point it's probably better to just render on the CPU.

Differential Revision: https://developer.blender.org/D2056

Cycles: make CUDA code a bit more robust to host/device alloc failures.

Fixes a few corner cases found while stress testing host mapped memory.

Merge branch 'blender2.8' into greasepencil-object