This may be a regression as I didn't notice it before.
Medium-roughness values end up producing spikes in the IBL sampling where, to the best of my understanding, a smooth gradient should be.
To me it looks like there's some wrong sampling going on on some wrap areas where the convolution goes wrong.
Description of Issue
This may be a regression as I didn't notice it before. Medium-roughness values end up producing spikes in the IBL sampling where, to the best of my understanding, a smooth gradient should be.
To me it looks like there's some wrong sampling going on on some wrap areas where the convolution goes wrong.
Steps to Reproduce
Download MetalRoughSpheres.Needle.zip
open in usdview
See discontinuities (Edit: accidentally had some spheres selected so they are tinted yellow, my bad):
This issue doesn't happen in other viewers, e.g. Khronos sample viewer – the model is "MetalRoughSpheres":
System Information (OS, Hardware)
Windows 11, RTX3060
Package Versions
23.11