Timothy Lochner

Any graphics engineers out there who structure their pixel and vertex shaders together? Because the two have to link together, I find it's easier to write 1 file with two entry points (ie, vs_main(), fs_main()) and then your UBOs and resources always match.

I *had* been doing this forever, but I've noticed that more and more, I'm decoupling geometry from shading and don't recommend this practice anymore. Especially as mesh-shaders come in, I think the two phases are separable and should be "mix-and-matched" more easily

I haven’t used mesh shaders yet. In what way do they impact the design choice, aside from being another stage in the pipeline?

Hmm mainly, I've been finding the cross-section of what's being shared from stage to stage is shrinking. Many studios I've worked with have had thousand+ lines of code in a single shader file, so I've found that just separating them (grouping within directories) helps my sanity