Cleaning Up Cartoons

[pulsar8]

Hey,

So I'm learning still but I do have the ability to do basic QTGMC encoding, applying filters, and upscaling, muxing etc.

What I want to do now is clean this image up from a DVD I own.

I think it needs a dotcrawl pass, and the color is mottled. There's ugly patterning on the colors etc.

Any suggestions for filters to apply? Do you agree it has dotcrawl? How would you characterize all the issues?

Thanks


[Attachment 71705 - Click to enlarge]

[pulsar8]

Here's a bit more information from DGINDEX.


[Attachment 71706 - Click to enlarge]

If it's interlaced then QTGMC should be useful and I don't need to do a inverse telecine?

[davexnet]

Here's a bit more information from DGINDEX.



If it's interlaced then QTGMC should be useful and I don't need to do a inverse telecine?

Post a minimum 10 second clip straight from the source (m2v cut from DGindex is good)
Something showing steady movement if possible

[pulsar8]

Okay so I figured this much out:

Deblock(quant=25, aOffset=25, bOffset=25, planes="yuv")

Before:


[Attachment 71707 - Click to enlarge]

After:


[Attachment 71708 - Click to enlarge]


So to my eyes it looks like deblocking gives a major boost to this source.

[pulsar8]

Here is the m2v... just learned how to save this

https://forum.videohelp.com/attachment.php?attachmentid=71709&stc=1&d=1686810035

[pulsar8]

Maybe some anti-aliasing too? The deblocking is the best thing I've done to it so far. I wonder why it got so blocky to begin with bad tape transfer? Thanks

[themaster1]

You can try this, though cartoons are not my forté

https://imgsli.com/MTg2MTA1

[pulsar8]

You can try this, though cartoons are not my forté

https://imgsli.com/MTg2MTA1

That's quite an improvement thanks. I'll have to study all these settings in this file. Lots to learn.

[jagabo]

Generally, telecined video should be inverse telecined, not deinterlaced.

The video isn't blocky. Using a deblocking filter removes some of the noise but only as a byproduct of it's processing. Dedicated spacial and temporal noise reduction filters will work better.

Halo reduction will help.

Mpeg2Source("sample.demuxed.d2v", CPU2="ooooxx", Info=3)  # enable deringing

Crop(8,0,-8,-0)
TFM(d2v="sample.demuxed.d2v") 
TDecimate() 

dehalo_alpha(rx=3.0, ry=2.0)

# blur away residual dot crawl artifacts near horizontal edges
hzlines = mt_edge("-16, -16, -16   16 16 16    0 0 0").mt_expand().Blur(1.5)
hzblurred = BilinearResize(288,height).Spline36Resize(width, height)
Overlay(last, hzblurred, mask=hzlines)

MCTemporalDenoise(settings="high")

aWarpSharp2(depth=5)
Sharpen(0.4, 0.2)

[Selur]

Another approach: Soft Telecine handling through DGDecNV + SCUNet: (don't think there's a SCUNet for Avisynth)

# Imports
import vapoursynth as vs
import os
import sys
# getting Vapoursynth core
core = vs.core
# Import scripts folder
scriptPath = 'F:/Hybrid/64bit/vsscripts'
sys.path.insert(0, os.path.abspath(scriptPath))
# Loading Plugins
core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/SourceFilter/DGDecNV/DGDecodeNV.dll")
# source: 'C:\Users\Selur\Desktop\sample.demuxed.m2v'
# current color space: YUV420P8, bit depth: 8, resolution: 720x480, fps: 29.97, color matrix: 470bg, yuv luminance scale: limited, scanorder: telecine (soft)
# Loading C:\Users\Selur\Desktop\sample.demuxed.m2v using DGSource
# using 'softpulldown' through DGDecNV
clip = core.dgdecodenv.DGSource("J:/tmp/m2v_d35cd4c630b0ddc4ebee73a92a6f76c2_853323747.dgi",fieldop=1)# 23.976 fps, scanorder: progressive
# Making sure content is preceived as frame based
# Setting detected color matrix (470bg).
clip = core.std.SetFrameProps(clip, _Matrix=5)
# Setting color transfer info (470bg), when it is not set
clip = clip if not core.text.FrameProps(clip,'_Transfer') else core.std.SetFrameProps(clip, _Transfer=5)
# Setting color primaries info (), when it is not set
clip = clip if not core.text.FrameProps(clip,'_Primaries') else core.std.SetFrameProps(clip, _Primaries=5)
# Setting color range to TV (limited) range.
clip = core.std.SetFrameProp(clip=clip, prop="_ColorRange", intval=1)
# making sure frame rate is set to 23.976
clip = core.std.AssumeFPS(clip=clip, fpsnum=24000, fpsden=1001)
clip = core.std.SetFrameProp(clip=clip, prop="_FieldBased", intval=0) # progressive
from vsscunet import scunet as SCUNet
# adjusting color space from YUV420P8 to RGBS for vsSCUNet
clip = core.resize.Bicubic(clip=clip, format=vs.RGBS, matrix_in_s="470bg", range_s="limited")
# denoising using SCUNet
clip = SCUNet(clip=clip, model=4)
# adjusting output color from: RGBS to YUV420P8 for x264Model
clip = core.resize.Bicubic(clip=clip, format=vs.YUV420P8, matrix_s="470bg", range_s="limited", dither_type="error_diffusion")
# set output frame rate to 23.976fps (progressive)
clip = core.std.AssumeFPS(clip=clip, fpsnum=24000, fpsden=1001)
# Output
clip.set_output()

Cu Selur

[pulsar8]

Another approach: Soft Telecine handling through DGDecNV + SCUNet: (don't think there's a SCUNet for Avisynth)

# Imports
import vapoursynth as vs
import os
import sys
# getting Vapoursynth core
core = vs.core
# Import scripts folder
scriptPath = 'F:/Hybrid/64bit/vsscripts'
sys.path.insert(0, os.path.abspath(scriptPath))
# Loading Plugins
core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/SourceFilter/DGDecNV/DGDecodeNV.dll")
# source: 'C:\Users\Selur\Desktop\sample.demuxed.m2v'
# current color space: YUV420P8, bit depth: 8, resolution: 720x480, fps: 29.97, color matrix: 470bg, yuv luminance scale: limited, scanorder: telecine (soft)
# Loading C:\Users\Selur\Desktop\sample.demuxed.m2v using DGSource
# using 'softpulldown' through DGDecNV
clip = core.dgdecodenv.DGSource("J:/tmp/m2v_d35cd4c630b0ddc4ebee73a92a6f76c2_853323747.dgi",fieldop=1)# 23.976 fps, scanorder: progressive
# Making sure content is preceived as frame based
# Setting detected color matrix (470bg).
clip = core.std.SetFrameProps(clip, _Matrix=5)
# Setting color transfer info (470bg), when it is not set
clip = clip if not core.text.FrameProps(clip,'_Transfer') else core.std.SetFrameProps(clip, _Transfer=5)
# Setting color primaries info (), when it is not set
clip = clip if not core.text.FrameProps(clip,'_Primaries') else core.std.SetFrameProps(clip, _Primaries=5)
# Setting color range to TV (limited) range.
clip = core.std.SetFrameProp(clip=clip, prop="_ColorRange", intval=1)
# making sure frame rate is set to 23.976
clip = core.std.AssumeFPS(clip=clip, fpsnum=24000, fpsden=1001)
clip = core.std.SetFrameProp(clip=clip, prop="_FieldBased", intval=0) # progressive
from vsscunet import scunet as SCUNet
# adjusting color space from YUV420P8 to RGBS for vsSCUNet
clip = core.resize.Bicubic(clip=clip, format=vs.RGBS, matrix_in_s="470bg", range_s="limited")
# denoising using SCUNet
clip = SCUNet(clip=clip, model=4)
# adjusting output color from: RGBS to YUV420P8 for x264Model
clip = core.resize.Bicubic(clip=clip, format=vs.YUV420P8, matrix_s="470bg", range_s="limited", dither_type="error_diffusion")
# set output frame rate to 23.976fps (progressive)
clip = core.std.AssumeFPS(clip=clip, fpsnum=24000, fpsden=1001)
# Output
clip.set_output()

Cu Selur

That's quite a result. It almost looks like it went through an AI upscale but obviously it did not.

I'll have to study this. Thanks.

[Selur]

Attached a few examples on what machine learning and anime specific upscalers look like. (no additional denoising aside from what the resizers offer done)

Cu Selur