hi,
in my own DX12 engine i am rendering completely bindless.
So very big meshes with many materials are drawn with one draw call
- using Vertex-, IndexBuffer ( fix )
- a additional material structered buffer (updated if needed)
- by indexing in a big texture array. (fix)
A flag in the material buffer tells if a material, ( defined by a sequenz of indexes in the index buffer ) shall be rendered or not.
Culling is done on CPU side by a bounding sphere tree.
What is the fastest way to do this ?
Do i need indirect rendering ?
Do i have to copy the indexes on GPU memory in a new buffer before submitting ?
I hoped there is a technique or GPU indirect function that i can feed on the GPU side that allows me
to say : at position x in indexbuffer take y count of indexed.
Waiting for your help, thanks