What is the proper way to ensure a render target is ready to be read by the cpu? I have a target that I render to, then generate mips with RHICmdList.GenerateMips
. Follow it with a render command fence ( RenderTargetForTextureTypeFence.BeginFence(true/*bSyncToRHIAndGPU*/)
) that a task thread waits on before locking the texture to use on cpu. I sometimes get no data (black pixels) on every mip except the largest.
We are on d3d11 on PC.