Dogway's Filters Packs

[kedautinh12]

Hi, sorry for question isn't relate post. I want prefilter convertoyuv444 but meet error: "both clip should same colorspace". I want to know what code can auto convert comeback original colorspace input clip?? Dogway

[Dogway]

What filter? SMDegrain? I tested and didn't get an issue. Post your call.


EDIT: By the way, happy not-so-happy day. Finally managed to craft a proper "optimized" sorting network for median7!! 6 months in the making finally. Well, actually it only saved 2 'max', 2 'dup1' and little else. Smart positioning of the lines has a greater impact on performance/multithreading. In any case speed is same as before since I have been feeling a slight slowdown of AVS+ from test20s.

[kedautinh12]

Oh, i asked a question for a code can convert comeback original colorspace based your prefilter but i know how to do that, thanks. Btw, i used to want give you a private message but your inbox full and i decide comment in here

[Dogway]

Yes, it's full many years 'round here. I have very "interesting" PM so I don't want to delete them ^^
I made MatchClip() to match format from clip 'a' to 'b'. In fast mode (default) it should work fine, my goal is to finish ConvertFormat() and default to it at some point.
You also have MatchColorFormat() from Raffriff42, and yesterday discovered HistogramRGB.avsi where it includes ConvertToMatch().

[StainlessS]

Yes, it's full many years 'round here. I have very "interesting" PM so I don't want to delete them ^^

You can save your PM's (to a file, I think - I've never done it).

[And just think of the very interesting PM's that you never got]

[kedautinh12]

Originally Posted by Dogway

Yes, it's full many years 'round here. I have very "interesting" PM so I don't want to delete them ^^
I made MatchClip() to match format from clip 'a' to 'b'. In fast mode (default) it should work fine, my goal is to finish ConvertFormat() and default to it at some point.
You also have MatchColorFormat() from Raffriff42, and yesterday discovered HistogramRGB.avsi where it includes ConvertToMatch().

And can you give me any posible contact when your massage full??

Originally Posted by Dogway

Yes, it's full many years 'round here. I have very "interesting" PM so I don't want to delete them ^^
I made MatchClip() to match format from clip 'a' to 'b'. In fast mode (default) it should work fine, my goal is to finish ConvertFormat() and default to it at some point.
You also have MatchColorFormat() from Raffriff42, and yesterday discovered HistogramRGB.avsi where it includes ConvertToMatch().

Hi Dogway,

Yes, you can indeed save your PM's (all at once), to either a .csv, .xml or .txt

Go to your PM page, select your PM's (top right hand corner, check box), then bottom right hand corner "Selected Messages", drop down to show option menu.

Then you can start again, with new "interesting" PM, as StainlessS mentioned.

[kedautinh12]

I met error with match clip:

ConvertToYV411: only 8 bit sources allowed
(C:/Program Files (x86)/AviSynth+/plugins64+/TransformsPack.avsi, line 144)
(C:/Program Files (x86)/AviSynth+/plugins64+/TransformsPack.avsi, line 145)
(C:/Program Files (x86)/AviSynth+/plugins64+/TemporalDegrain-v2.4.1.avsi, line 440)
(C:/Program Files (x86)/AviSynth+/plugins64+/TemporalDegrain-v2.4.1.avsi, line 345)
(C:\Megui\AVSMeter3090\td-002.mkv.avs, line 4)

can you tell me any error with this function??

function DT_BM3D(clip a, float "sigma", int "radius", bool "CUDA", bool "chroma", int "device_id") {

s = Default(sigma, 3)
r = Default(radius, 1)
cd = Default(CUDA, true)
chr = Default(chroma, false)
device_id = Default(device_id, 0)
r = min(r,3) # Buggy when r > 3
a
ch = ((chr == true) && !Is444(a))
ch ? ConverttoYUV444(chromaresample="spline16") : last
cd ? BM3D_CUDA(sigma=s, radius=r, chroma=chr, fast=true, extractor_exp=6, device_id=device_id) : \
BM3D_CPU (sigma=s, radius=r, chroma=chr)
BM3D_VAggregate(radius=r)
ConvertBits(16, dither=1)
ch ? MatchClip(a) : mergechroma(a)
}

[Dogway]

Thx, yes PMs when Doom9 was an hostile place. I like to keep a record of internet beefs as fun facts ^^
I will do backup then to start anew for the new year.

@kedautinh12: Are you testing with 411 clip? AVS+ doesn't support HBD 411, and BM3D doesn't support anything less than 32-bit.

You can fix the function with ConvertBits(BitsPerComponent(a)) before MatchClip(a), I need to implement range and bitdepth matching into MatchClip() (already done), will update soon.

[kedautinh12]

Originally Posted by Dogway

Thx, yes PMs when Doom9 was an hostile place. I like to keep a record of internet beefs as fun facts ^^
I will do backup then to start anew for the new year.

@kedautinh12: Are you testing with 411 clip? AVS+ doesn't support HBD 411, and BM3D doesn't support anything less than 32-bit.

You can fix the function with ConvertBits(BitsPerComponent(a)) before MatchClip(a), I need to implement range and bitdepth matching into MatchClip() (already done), will update soon.

I'm testing with 420 clip

edit: i added ConvertBits(BitsPerComponent(a)) but error still exist. I don't know why my testing clip is 420 but when use with MatchClip(a), it's become 411

[Dogway]

Can you show a small script showing the problem?

[kedautinh12]

script showing problem

ConvertToYV411: only 8 bit sources allowed
(C:/Program Files (x86)/AviSynth+/plugins64+/TransformsPack.avsi, line 144)
(C:/Program Files (x86)/AviSynth+/plugins64+/TransformsPack.avsi, line 145)
(C:/Program Files (x86)/AviSynth+/plugins64+/TemporalDegrain-v2.4.1.avsi, line 442)
(C:/Program Files (x86)/AviSynth+/plugins64+/TemporalDegrain-v2.4.1.avsi, line 345)
(C:\Megui\AVSMeter3090\td-002.mkv.avs, line 4)

here my script

function DT_BM3D(clip a, float "sigma", int "radius", bool "CUDA", bool "chroma", int "device_id") {

bi = BitsPerComponent(a)
s = Default(sigma, 3)
r = Default(radius, 1)
cd = Default(CUDA, true)
chr = Default(chroma, false)
devID = Default(device_id, 0)
r = min(r,3) # Buggy when r > 3
a
ch = ((chr == true) && !Is444(a))
ch ? ConverttoYUV444(chromaresample="spline16") : last
cd ? BM3D_CUDA(sigma=s, radius=r, chroma=chr, fast=true, extractor_exp=6, device_id=devID) : \
BM3D_CPU (sigma=s, radius=r, chroma=chr)
BM3D_VAggregate(radius=r)
ConvertBits(16, dither=1)
ConvertBits(bi)
ch ? MatchClip(a) : mergechroma(a)
}

[Dogway]

That's not a script but a function declaration. Anyway I updated TransformsPack so check if it fixes your problem.

I'm going to rest for a month or so now, I will still be uploading updates but less frequently.

[kedautinh12]

i updated and still error. my script:

SetFilterMTMode("DEFAULT_MT_MODE", MT_MULTI_INSTANCE)
LoadPlugin("C:\Megui\MeGUI-2924-64\tools\lsmash\LSMASHSource.dll")
LWLibavVideoSource("C:\Megui\td-002.mkv")
TemporalDegrain2(postFFT=5)

TemporalDegrain test ver:
https://drive.google.com/file/d/1V3x...ew?usp=sharing

[Dogway]

It wasn't consistent with bitdepth. You converted to 32-bit before DT_BM3D, so afterwards DT_ContraSharpening was handling one clip in 16-bit and another in 32-bit.

  
  #---------------------------------------
  # post FFT

  fullClip = dgNR2
  
  fullSuper = (postTR > 0) ? fullClip.MSuper( pel=subpel, levels=1, chroma=ChromaNoise ) : NOP() 
  noiseWindow = (postTR == 0) ? fullClip :				\
                fullClip.MCompensate( fullSuper, vmulti.SelectRangeEvery(maxTR*2,postTR*2), tr=postTR, center=true, thSAD=thSAD2, thSCD1=thSCD1, thSCD2=thSCD2 )
                           
  dftDither = (postDither < 0) ? postDither*(-1) : 0
  postDither = (postFFT == 5 && postDither < 1) ? 1 : postDither  # ensure to ConvertBits down after BM3D
  dnWindow = (postFFT == 0) ? RemoveGrain( postDither > 1 ? noiseWindow.ConvertBits(16) : noiseWindow, mode=1) : \
             (postFFT == 1) ? neo_fft3d ( postDither > 0 ? noiseWindow.ConvertBits(16) : noiseWindow, y=LumaNoise?3:2, u=ChromaNoise?3:2, v=ChromaNoise?3:2, sigma=postSigma, bt=postTD, ncpu=fftThreads ) : \
             (postFFT == 11)? FFT3DFilter( postDither > 0 ? noiseWindow.ConvertBits(16) : noiseWindow, plane=degrainPlane, sigma=postSigma, bt=postTD, ncpu=fftThreads ) : \
             (postFFT == 2) ? FFT3DGPU( postDither > 0 ? noiseWindow.ConvertBits(16) : noiseWindow, plane=degrainPlane, sigma=postSigma*2/3, bt=postTD, precision=2, mode=1 ) : \
             (postFFT == 3) ? neo_dfttest( postDither > 0 ? noiseWindow.ConvertBits(16) : noiseWindow, y=LumaNoise?3:2, u=ChromaNoise?3:2, v=ChromaNoise?3:2, sigma=postSigma*4, tbsize=postTD, dither=dftDither, threads=fftThreads ) : \
             (postFFT == 13)? dfttest( postDither > 0 ? noiseWindow.ConvertBits(16) : noiseWindow, Y=LumaNoise, U=ChromaNoise, V=ChromaNoise, sigma=postSigma*4, tbsize=postTD, threads=fftThreads, dither=dftDither ) : \
             (postFFT == 4) ? DT_KNLMeansCL( postDither > 0 ? noiseWindow.ConvertBits(16) : noiseWindow, a=2, d=postTR, h=postSigma, Luma = LumaNoise, Chroma = ChromaNoise, device_type="GPU", device_id=devId) : \
             (postFFT == 5) ? DT_BM3D( postDither > 0 ? noiseWindow.ConvertBits(16) : noiseWindow, radius=postTR, sigma=postSigma, CUDA=cuda, chroma=ChromaNoise, device_id=devId ) : \
             (postFFT == -1)? HQDn3D( postDither > 0 ? noiseWindow.ConvertBits(16) : noiseWindow, 0,0,4,1, u=ChromaNoise?3:2, v=ChromaNoise?3:2) : NOP()
  dnWindow = dnWindow.SelectEvery( postTD, postTR )
  denoised = (postDither == 1) ? dnWindow.ConvertBits(outputBits,dither=1) : dnWindow

  csOrg = (postDither == 2) ? dgO.ConvertBits(16) : dgO
  sharpened = DT_ContraSharpening(denoised, csOrg, extraSharp)
  sharpened = (postDither == 2) ? sharpened.ConvertBits(outputBits,dither=1) : sharpened
  
  # mix with original to leave the posibility to decrease the denoising strength
  sharpened = postMix > 0 ? mt_lutxy(dgO,sharpened,"x "+String(postMix)+" * y "+String(int(100-postMix))+" * + 100 /") : sharpened

  # Crop off temporary vertical padding
  cropped = sharpened
  
  output = debug ? Select(outputStage, dgNR1x, dgNR2, cropped, repair0, spatialBlur, srchClip, dgLimit, denoised) : Select(outputStage, dgNR1x, dgNR2, cropped)
  return output
}


function DT_BM3D(clip a, float "sigma", int "radius", bool "CUDA", bool "chroma", int "device_id") {

    bi = BitsPerComponent(a)
    s     = Default(sigma,      3)
    r     = Default(radius,     1)
    cd    = Default(CUDA,    true)
	chr   = Default(chroma, false)
	devID = Default(device_id,  0)
    r     = min(r,3) # Buggy when r > 3
    a
	ch = ((chr == true) && !Is444(a))
    ch ? ConverttoYUV444(chromaresample="spline16").ConvertBits(32) : ConvertBits(32)
    cd ? BM3D_CUDA(sigma=s, radius=r, chroma=chr, fast=true, extractor_exp=6, device_id=devID) : \
         BM3D_CPU (sigma=s, radius=r, chroma=chr)
    BM3D_VAggregate(radius=r)
	ConvertBits(bi, dither=1)
    ch ? MatchClip(a) : mergechroma(a)
}

[Dogway]

By the way, last night I had an enlightenment and today I refactored ex_vibrance(). Fast, HBD support and straight to the point. 'Recover' mode also works now to recover lost saturation by filtering, derainbows, etc. I plan to revisit SkinQualifier soon.

[anton_foy]

Originally Posted by Dogway

By the way, last night I had an enlightenment and today I refactored ex_vibrance(). Fast, HBD support and straight to the point. 'Recover' mode also works now to recover lost saturation by filtering

That sounds amazing just what I need!

[Dogway]

Yes, I remember when you needed something similar a few months back. Recover mode never was implemented so this is some achievement, actually it was very easy, just divide the masks but I wasn't so math savvy then.

[Fjord]

Fantastic collection of functions with references!
Just browsing your code - noted a string with trailing space "CAM16 " at line 2822 in function Format_fuzzy_search(...)
...
cm=
...
\ : (FindStr(model, "CAM16" )>0) ? ["CAM16 ", "YUV",34]

I assume the trailing space is unintentional.

[Dogway]

Yes, thanks! Anyway many models aren't implemented yet, CAM16 is said to be complex while not being so uniform as recent ones. TransformsPack (namely ConvertFormat()) is a long term project, the rest are pretty mature.

I updated GradePack with some optimizations for ex_vibrance() and a new filter, Skin_Qualifier() to create skin masks. It works like raffriff42's SmoothHueMask() with different false positives, but generally less amount of them. It's also slower since it's processed in HBD internally and a conversion to OkLab is performed for perceptually uniform HUE. Runs at about 60fps on my CPU -Prefetch(8)-

I'm going now to implement HSVxHSV, that is; HUExHUE, HUExSAT, SATxLUM, etc.