
# Example: last.FFT3DFilter(). # In this mode, you can safely set Lmode=0 & PP=off QTGMC is used here with Vapoursynth, as you can see in the videos themselves. # If source is defined, LSFmod doesn't sharp more a denoised clip than this source clip Some examples of the use of QTGMC deinterlacer. # Mode to avoid banding & oil painting (or face wax) effect of sharpening The lowerer the MDSI score, the higher the image perceptual quality. MDSI (Mean Deviation Similarity Index) is a full reference IQA model that utilize gradient similarity (GS), chromaticity similarity (CS), and deviation pooling (DP). # Mode to avoid noise sharpening & ringing Use MXNet to accelerated Image-Processing in VapourSynth using a C++ interface. # =2 : Nonlinear sharpening (corrected version) # - added warning about bad settings (no more silent) My main interest in VS is that it has a native QTGMC port, and before AVISynth+ 64-bit got stable, VS has been a preferred method among some users for faster/more stable conversions. # - added new calculation for soft (soft=-2) For those who don't know, VapourSynth is a python-based video processing scripting system similar to AVISynth, and can actually use AVISynth plugins. # v1.3 : - changed a little Smethod=3&5 (same effect, but more precise) # v1.4 : - changed defaults="new" to defaults="slow" & defaults="fast" # v1.5 : - fixed LUT expression (thanks to Didée) # v1.8 : - changed preblur to allow more tweaking (bool->string) # - default preset is now defaults="fast" /!\ Basically a debander with as many of the perks of a bilateral filter that it can afford, and to me Dither_bilateral16 fits the bill very well.# LimitedSharpenFaster MOD : function LSFmod() #

In this case, you want a controlled smoothing of the flat areas with a wide enough distribution to reconstruct gradients, without taking higher detail into account.

VS Bilateral works fine for light banding (probably better), but not if it's a source where say, it was post-processed in 8 bit, high depth was rounded rather than dithered, or it's just plain bitstarved. All the more so when the banding is moving It being a good bilateral works to its disadvantage when you need to actually reconstruct most of the gradients. It just sorta softens them, without distributing the color to a wider area, which doesn't seem to change much with stronger sigmaR settings. It doesn't smooth as much along more pronounced banding edges.
