Since the GPU occlusion culling is computed from the depth texture of the current and the last frame, it would be very useful to query the occlusion function from a compute shader. For example, I have a compute shader that selects & culls instances to be rendered using Graphics.DrawMeshInstancedIndirect.
Is that currently possible and if not, is it going to be?
If the current implementation is something like Hi-Z occlusion culling we can surely just access the Hi-Z buffer?
Assuming you’re talking about GPU Resident Drawer Occlusion Culling, it just outputs to the same buffer as the depth prepass. So you should be able to access it via the normal render pipeline API.
Occlusion functions though (so basically what’s in Runtime\GPUDriven\OcclusionCullingCommon.hlsl) are not “exposed”. They are not really usable outside of the occlusion culling passes that are issued by GPU Resident Drawer. This is because they depend on various parameters stored in a cbuffer managed by GRD that’s only bound during those passes.
Perhaps it would be possible to expose those by making the cbuffer global or something. But that’s not the case currently.