Blender Git Loki

Cleanup: fix various warnings when building CUDA/OptiX kernels

Cycles X: Use approximate shadow catcher in viewport

This change makes it so that shadow catcher objects have effect in
the viewport. Currently is limited to the approximate mode since it
is the only one possible with the draw manager integrator.

Differential Revision: https://developer.blender.org/D11484

EEVEE: LightProbe: Fix grid display and deduplicate code

EEVEE: LightProbe: Fix light cache display visibility without overlays

EEVEE: LightProbe: Fix wrong shading on other cube faces

Use modified viewmat instead of modified winmat. Too much code
assume conventional winmat.

EEVEE: LightProbe: Fix level bias and visibility filtering

EEVEE: Fix irradiance baking sample order

Cycles X: Allow viewing shadow catcher pass in viewport

Differential Revision: https://developer.blender.org/D11482

Cycles X: Implement pass accessor for GPU

It replaces film_convert_half_float with more generic code path.

Currently only half4 destination is supported. No real stopper reason
to support float destination, is just a bit harder to verify at this
point as there will be no users of this code path.

Form quick benchmarks the new code seems to be 10-15% faster than the
film_convert_half_float when measuring overall display update time for
non-interop rendering. Partially because we no longer convert global
index to x/y and back to pixel index, partially due to smaller kernel.

Note that the timing is only taking display update into account, the
path tracing is not included into the speedup calculation (this means
that viewport is not that much faster, but it means we can update
viewpor4t more often now).

Differential Revision: https://developer.blender.org/D11481

Cycles X: Use pass accessor for viewport on CPU

No functional changes, no measurable time difference.

Quite boring on itself, but it opens doors for support of shadow
catcher pass in the viewport.

Differential Revision: https://developer.blender.org/D11480

Cycles X: Multi-thread CPU pass accessor

Is not very measurable for the combined pass, but for shadow catcher
matte with approximated shadow gives about 30% speedup of the display
update.

Surely, this is only fraction of the overall render time, but is
still nice to have it. Makes it easier to re-use for the viewport
update as well.

Differential Revision: https://developer.blender.org/D11479

Cycles X: Move pass pixel processors to kernel_film

Currently no functional changes, just keep moving towards unifying
calculation logic between CPU and GPU.

There combined pass is equal in timing within a noise level, the
shadow catcher matte pass with approximated shadow is about 7%
faster (probably, due to difference in inline policy).

Differential Revision: https://developer.blender.org/D11478

Cycles X: Move top-level pass access log to a base class

What this change allows to do is to share top-level logic of pixel
access between various implementations (CPU, GPU), while keeping the
actual pixel processing very small.

Currently there are no functional changes, as well no measurable
difference in timing spent in getting tile pixels.

Differential Revision: https://developer.blender.org/D11477

Cycles X: Move pass accessor to integrator folder

This is where it is used the most and will be used even more in the
future.

No functional changes, preparing for an upcoming generalization of
the class. Eventually this class will take over the kernel film
convert kernel, for which DeviceQueue will be needed, so it makes
sense to move the pass accessor to the integrator folder

Differential Revision: https://developer.blender.org/D11476

EEVEE: Fix irradiance smoothing wrong

EEVEE: LightCache: Port baking

Some things differs from old implementation.
- Object visibility is filtered correctly without using a visibility
callback (which is to be removed).

The implementation is also more high level using less low level tricks.

A dedicated LightProbeView is created for each lightprobe cubeface to
render using all pipeline (deferred and forward).

There is still a few things not working.

BLI: float3: add division operator between 2 float3

EEVEE: LightCache: Port back cache Display

Nothing significantly different appart from codestyle.

EEVEE: Fix background transparent

Alpha of main output is transparency and not opacity.

Cycles X: Better handling of missing CUDA kernel functions

Log missing function and avoid crash.

Differential Revision: https://developer.blender.org/D11472