Open bagusman7 opened 1 year ago
Hmm, this looks like a singularity (a 0div). Not sure where this would be since a) this seems to be inconsistent between GPUs (let me guess: you have a Nvidia card?) and b) there aren't many strong candidates for this in the code.
Thanks for the quick response, and yes, I'm running this off a 1660s.
Confirm the same here on a 3060. I tried a pixel history debug through RenderDoc but OpenMW's GL context is so old that RD doesn't support it. =(
Confirm the same on RTX3050. What solved it for me was in my NVIDIA control panel sessions I had to separately turn on Antialiasing-FXAA. Hope this helps.
Still happening with openmw 0.49 (also nvidia). Forcing fxaa in the driver makes no difference. OS is windows 10.
FXAA might detect the artifacts as edges in the luminosity phase and blend them. So for some users it will appear to fix the problem, but it's literally a band-aid.
Turning anisotropic filtering to application controlled in NVIDIA settings seemed to have solved the issue for me.
Oh, that's interesting... Perhaps the GPU is getting the wrong end of the stick when it comes to calculating deltas for filtering? I might look into that if I get time.
Had this issue as well with a 3080 in Linux Mint. Was able to fix by both enabling FXAA in the NVIDIA Settings app, as well as disabling Override Application Settings for Anisotropic Filtering. No clue why that was enabled...
Oh, disabling anisotropic filtering did the job? I'm starting to wonder whether this is an issue with Nvidia's anisotopic filtering when combined with tiling... Regardless, anisotropic filtering shouldn't be enabled for this use-case.
Sorry for my lack of technical terminology, but I'm getting these strange lines across the clouds, only when this shader is active. I'm using OpenMW dev build 613453c799. loadorder.txt