Jaded-Encoding-Thaumaturgy/vs-denoise

Fluxsmooth params should be scaled based on bit depth.

Closed this issue · 0 comments

temp_thr, spat_thr = kwargs.get('temp_thr', 2), kwargs.get('spat_thr', 2)
return min_blur(clip, 2, planes).flux.SmoothST(temp_thr, spat_thr, planes) # type: ignore

I literally just ran into this issue, where the fluxsmooth plugin appears useless on high bit depth (16) content. Turns out there's a known "issue" on this: dubhater/vapoursynth-fluxsmooth#3

Looks like simply scaling the threshold params based on bit depth works fine in my testing.