Arseny Kapoulkine

The scene above has the same issue on gltfpack 0.16 as far as I can tell. This came up before (#433) and is due to some odd export behavior that...

Note: for any report in this thread that doesn't come with a link to the scene before the compression, it's going to be impossible to investigate whether it's something that...

Since no new examples have been posted, and https://github.com/zeux/meshoptimizer/issues/466#issuecomment-1221746038 is inevitable / expected, I'm going to assume that there's no regression and that the only tangible issue is with the...

Yeah, `.c_str()` is already the correct way to handle this. My understanding is that it's not really legal to forward-declare `std::filesystem::path` (and, in practical terms, pugixml used to try to...

Yeah that makes sense. I believe it's also the case that writing metadata might fail during `fclose`, so ideally we should check the result of `fclose` instead.

One issue with this is that all of this requires rebuilding meshlet data, something that would ideally be done offline. It looks as if on AMD hardware specifically the shader...

I'm also wondering what happens on NV specifically from work groups with size 64 vs 32 - and whether the shader that the driver runs is substantially different wrt performance...

> No, it doesn't require rebuilding meshlet data. A workable compromise is if you use a meshlet size of 64 (max 64 vertices and max 64 primitives). In this case...

Can you elaborate on why on AMD there's a benefit to going above 64? It's not intuitively obvious that this should help as 64 (and sometimes 32) is the HW...

Ah, that explains a lot! It's indeed substantially different compared to NV model. I didn't realize that the restriction on emission also applies to primitives, as I thought it's just...