There are 6 index buffers in FStaticMeshLODResources and I’d like to clarify their purpose. 4 seem to be the same size and are rearrangements of each other.
IndexBuffer is required for rendering and AdjacencyIndexBuffer is required for tessellated rendering. The wireframe buffer is editor only, so not an issue. These are good to go =)
I presume the depth only index buffer is for a depth only pass to support meshes with translucent materials. As the mesh batching is done on a material/section basis, this would be done automatically as part of the batching process using the regular index buffer?
The reverse index buffers seem to prevent a single state change per material if there is a mix of models with the same material where some are inside out and some are not (a mixture of signs the determinant of the transform). Is this worth the extra memory now we have DX12/Vulkan/Metal?
We are looking to add multiple versions of the same LOD with different index buffers, so cutting this set of 6 down would be great!
Cheers
John