I’m currently working on a project and I’m curious to know if it’s possible to implement a multi-pass post-process without utilizing a global shader. My goal is to use the output from one post-process material as an input for the next material in the pipeline.
Is this achievable? If so, any tips or examples would be greatly appreciated.
it’s ill advised to do a second full screenpass. under normal conditions you should merge it into one shader. you still get 2 passes. just in one shader.
you only do multiple passes if you have a large sample kernel. where you should do a horizontal and vertical shader to avoid to much gpu cache clutter.
I’m facing a situation where the first pass involves sampling surrounding pixels for computation, and the next pass requires blurring the results.
I attempted merging the blur into the first pass, but it’s too resource-intensive. Any suggestions?