avisynth plugin requests

[edcrfv94]

I can't find any specialized thread request plugins for VapourSynth, so I just start a new thread, hope it's not break forum rules.

warpsharp(only need WarpSharp)
Nice Anime sharpening filter.

TEdgeMask(dfmderainbow_mc requirement)
GenericFilters still not releases.

TCombMask
Remove combing core filter.

IsCombedTIVTC(vivtc.VFM with out matching)
example (https://forum.doom9.org/showthread.php?p=1745735#post1745735)

deint = mt_merge(last, QTGMC(FPSDivisor=2), TCombMask(lcLinked=true).RemoveGrain(4, -1).mt_expand(mode=mt_square(2)).mt_inflate(), luma=true)
ConditionalFilter(last, deint, last, "IsCombedTIVTC(cthresh=6, MI=64)", "=", "true")


TBilateral(DeHaloHmod requirement)
Also can remove residual combing with TCombMask+IsCombedTIVTC

[poisondeathray]

warpsharp(only need WarpSharp)
Nice Anime sharpening filter.


what about awarpsharp2 ? or is there a specific reason for warpsharp ?

https://github.com/dubhater/vapoursynth-awarpsharp2
https://forum.doom9.org/showthread.php?t=172721

[WolframRhodium]

IsCombedTIVTC - core.tdm.IsCombed
https://github.com/HomeOfVapourSynthEvolution/VapourSynth-TDeintMod

TBilateral - core.bilateral.Bilateral
https://github.com/HomeOfVapourSynthEvolution/VapourSynth-Bilateral

[edcrfv94]

what about awarpsharp2 ? or is there a specific reason for warpsharp ?

https://github.com/dubhater/vapoursynth-awarpsharp2
https://forum.doom9.org/showthread.php?t=172721

awarpsharp2 look like too thin to me, if slightly sharp warpsharp look like much better.

IsCombedTIVTC - core.tdm.IsCombed
https://github.com/HomeOfVapourSynthEvolution/VapourSynth-TDeintMod

TBilateral - core.bilateral.Bilateral
https://github.com/HomeOfVapourSynthEvolution/VapourSynth-Bilateral

Thank you remind me TDeintMod include IsCombed.
Bilateral the results quite different than TBilateral.

[ChaosKing]

awarpsharp2 look like too thin to me, if slightly sharp warpsharp look like much better.


depth
Controls how far to warp. Negative values warp in the other direction, i.e. will blur the image instead of sharpening.

Just set a lower depth value ;)

[lansing]

Awarpsharp should not be use as a sharpener directly on the video

[ChaosKing]

How would you use it? Can you give an example?
I use it directly with low depth values...

[lansing]

How would you use it? Can you give an example?
I use it directly with low depth values...

You don't use it directly on the video because it thins the lines, it destroys the original image. This filter can be use as part of other evaluation filters. For example, in avisynth, it was use within mergechroma to fix colorbleed.

[kgrabs]

Perhaps I can piggyback on this and request a Dither_bilateral16 port? For anything that really needs debanding, VapourSynth-Bilateral is pretty much useless (and lolslow to boot)
It's pretty much the sole reason I still use Avisynth for half of the shows I encode (the more-banded half, that is)

Also I have the dehalohmod mask ported, op: https://pastebin.com/raw/2uziEgxr

[Myrsloik]

Perhaps I can piggyback on this and request a Dither_bilateral16 port? For anything that really needs debanding, VapourSynth-Bilateral is pretty much useless (and lolslow to boot)
It's pretty much the sole reason I still use Avisynth for half of the shows I encode (the more-banded half, that is)

Also I have the dehalohmod mask ported, op: https://pastebin.com/raw/2uziEgxr

VapourSynth-Bilateral doesn't even produce similar results? Or the problem is just that it's slow?

[kgrabs]

It doesn't smooth as much along more pronounced banding edges. 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 being a good bilateral works to its disadvantage when you need to actually reconstruct most of the gradients. All the more so when the banding is moving

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. 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. 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.

e: I would be lying if I said the speed wasn't an issue though, since it feels like adding insult to injury, but I hope to nix that issue by getting a GTX 1060 and using the opencv version soon

[WolframRhodium]

e: I would be lying if I said the speed wasn't an issue though, since it feels like adding insult to injury, but I hope to nix that issue by getting a GTX 1060 and using the opencv version soon

I had ported OpenCV's Bilateral which uses CUDA at VapourSynth-BilateralGPU.

But to be honest, for pure debanding, I suggest you to use other fast algorithms like Guided Filter (https://github.com/WolframRhodium/muvsfunc/blob/ff4eba925e5688167d68fa18941a1a3a782e25c1/muvsfunc.py#L2820), SmoothGrad (https://github.com/WolframRhodium/muvsfunc/blob/ff4eba925e5688167d68fa18941a1a3a782e25c1/muvsfunc.py#L1821), or even just core.f3kdb.Deband (https://github.com/SAPikachu/flash3kyuu_deband) combined with the powerful technique named nr-deband (https://pastebin.com/7zB0qz0H) (proposed by the author of VapourSynth-Bilateral. Here (https://pastebin.com/NGFBhCCn) is the code in VS that illustrates the main idea).
Advanced techniques include masking, limiting, pre-denoising, post-sharpening, etc. That's another story.

[Myrsloik]

...

I completely missed that you have a pile of possibly useful scripts. I'll add them to the list.

[WolframRhodium]

I completely missed that you have a pile of possibly useful scripts. I'll add them to the list.

Thanks.

[kgrabs]

Well f3kdb and SmoothGrad aside, the Guided Filter looks pretty neat. I'll probably be trying that out very soon. I do recall mawen1250 discouraging anything other than gauss/average blurring for the f3kdb prefilter thing, idk how dfttest would turn out, seeing as it has a bit of a debanding effect on its own. It's not perfect, but it does do a pretty good job.

Edit: I'm still getting used to it, but so far it seems GuidedFilter on regulation mode 2 is plenty good enough to replace Dither_bilateral16

[masterkivat]

Is it possible to port "Strange.avsi (https://www.animemusicvideos.org/forum/viewtopic.php?t=107160#p1411250)" script from mirkosp to Vapoursynth? :confused:

[lansing]

Is it possible to port "Strange.avsi (https://www.animemusicvideos.org/forum/viewtopic.php?t=107160#p1411250)" script from mirkosp to Vapoursynth? :confused:
You can use ReplaceFrameSimple from RemapFrames (https://github.com/Irrational-Encoding-Wizardry/Vapoursynth-RemapFrames)

clip = core.ffms.source("yo.mp4")
nnedi3clip = core.nnedi3.nnedi3(clip)
clip = core.remap.Rfs(clip, nnedi3clip, mappings="[10 20] 25 30")

[Selur]

While at it: A Vapoursynth version of FrameRateConverter (https://forum.doom9.org/showthread.php?t=174793) would be cool. :)

[KingLir]

Also I have the dehalohmod mask ported, op: https://pastebin.com/raw/2uziEgxr

Hi, can you port the entire DeHaloHmod function as well ?

[zorr]

I'd like to request a port of Median / MedianBlend (https://forum.doom9.org/showthread.php?t=170216). Documentation here (http://avisynth.nl/index.php/Median#MedianBlend).

There's a version of "toot" already which ignores the worst of three but I want to take five clips, throw away the highest and lowest samples and take an average of the remaining three. I could do this in Avisynth but I also need to do a small brightness adjustment to the clips before the blending and I think doing the whole process in 10bits would help with the accuracy (Avisynth version doesn't support that I think).

[ChaosKing]

+1 MedianBlend is awesome. I once "cleaned" an series opening by combining all episodes this this plugin. Also avs version can't be loaded in vapoursynth, error is:
Avisynth Compat: varargs not implemented so I'm just gonna skip importing Median
Avisynth Compat: varargs not implemented so I'm just gonna skip importing MedianBlend

[zorr]

Has anyone made port of MatchClips (https://forum.doom9.org/showthread.php?t=86691) yet? It's useful for matching the clips temporally before applying the MedianBlend. This one could be implemented as a python script, at least most of the functionality.

[ChaosKing]

Nope, but source can be found here http://www.avisynth.nl/users/warpenterprises/

[sl1pkn07]

Also I have the dehalohmod mask ported, op: https://pastebin.com/raw/2uziEgxr


can you upload it to github or something?

greetings