8/19/2021
Got everything working for now.
Now trying to figure out the most elegant way to tie upates to the render thread.
Moving forward i will look into LODs, RDG graph, Async compute, might implement dual marching cube as well.
8/17/2021
Single draw call and updated every frame, still no async compute, WIP as there’s a bug i need to figure out still:
8/15/2021 Update:
I am working toward Compute Shader direct/indirect drawing, meaning computed data on the GPU memory is directly used for rendering without having to copy back data to the RAM for any CPU management.
Made a short demo to showcase the marching cube i just implemented.
Really fun to be able to generated meshes using distance fields!
What i have been doing initially was computing the data on the GPU reading it back/copying to the RAM for CPU usage and creating a standard procedural mesh from it.
Right now i’m trying to draw directly from the output data on the GPU without any (or minimal) readbacks. My bottleneck isn’t tied to those read backs mind you, but rather to the fact i’m not leveraging async compute for now.