Pink spots on VHS MESECAM Capture

[gleim]

Hi everybody! I'm from Russia, and this cassette was recorded in MESECAM. There is a problem - it has a lot of chroma noise, such as these pink spots:

[Attachment 47239 - Click to enlarge]

Is it possible to clean these spots? This video has another problem like chroma shifting, color bleeding - I corrected them as I could:

AVISource("D:\Pif_Sample.avi")
XviD4PSPPluginsPath = "C:\Program Files (x86)\XviD4PSP 5\dlls\AviSynth\plugins\"
LoadPlugin(XviD4PSPPluginsPath + "Warpsharp.dll")
Import(XviD4PSPPluginsPath + "RemoveColorBleeding.avs")
LoadPlugin(XviD4PSPPluginsPath + "fft3dfilter.dll")
converttoyv12()
ChromaShift(c=12)
MergeChroma(aWarpSharp(depth=20))
RemoveColorBleeding()
fft3dfilter(bw=32,bh=32,ow=16,oh=16,bt=2,sigma=3.0,sigma2=2.5,sigma3=16.0,sigma4=10.0,plane=3,ncpu=1)
Crop(12, 2, -16, -8)
maa2()

RemoveColorBleeding.avs:

# RemoveColorBleeding (by Didйe)
function RemoveColorBleeding(clip input) {
o=input
warp1=o.mergechroma(o.blur(1.0).unsharpmask(255,1,1).aWarpSharp2(depth=30, thresh=192, blur=3, chroma=4))
warp2=o.mergechroma(warp1.blur(1.0).unsharpmask(255,1,1).aWarpSharp2(depth=12, thresh=192, blur=2, chroma=4))
# interleave(o,warp1,warp2) # visualization
return warp2
}

I uploaded two samples - source and encode. I would be very grateful for any help.

[themaster1]

Try this, it works great with cartoons generally; require the filters: TTempSmooth, variable blur, KNLMeansCL, camcorder color denoise (vdub):

AVISource("Pif_Sample.avi")
assumetff()
###1) Clean Chroma artefacts (heavy cleaning):
ConvertToYV16(interlaced=true)
orig=last
ev=orig.assumetff().separatefields().selecteven()
od=orig.assumetff().separatefields().selectodd()
ev
ue_chroma = UToY(ev).blur(0,1.5).binomialblur(5).ttempsmooth(m axr=6,lthresh=150, strength=6).KNLMeansCL(d=3,a=8,s=2,h=6)
ve_chroma = VToY(ev).blur(0,1.5).binomialblur(5).ttempsmooth(m axr=6,lthresh=150, strength=6).KNLMeansCL(d=3,a=8,s=2,h=6)
YToUV(ue_chroma, ve_chroma)
MergeLuma(ev)
ev_filtered=last
od
uo_chroma = UToY(od).blur(0,1.5).binomialblur(5).ttempsmooth(m axr=6,lthresh=150, strength=6).KNLMeansCL(d=3,a=8,s=2,h=6)
vo_chroma = VToY(od).blur(0,1.5).binomialblur(5).ttempsmooth(m axr=6,lthresh=150, strength=6).KNLMeansCL(d=3,a=8,s=2,h=6)
YToUV(uo_chroma, vo_chroma)
MergeLuma(od)
od_filtered=last
interleave(ev_filtered,od_filtered)
assumefieldbased().assumetff().weave()
###2) Clean Chroma artefacts (soft cleaning in rgb with CCD filter):
ConverttoRGB32(matrix="rec601",interlaced=true)
separatefields()
LoadVirtualDubPlugin("C:\Program Files (x86)\virtualdubmod1.5\plugins\Camcorder_Color_Den oise_sse2.vdf", "CCD", 0)
CCD(10,1) # 0-100, default 30
weave()
converttoyv12(matrix="Rec601",interlaced=true)
## 3) remove some chroma bleeding:
mergechroma(awarpsharp2(depth=10, type=1, blur=1, chroma=4)) # chroma=4 ->luma guide

[themaster1]

filtered / original:

[gleim]

themaster1
Thank you so much! That is much better. But on the first frames pink spots still there. How can I set more stronger filtration? I tried to set more higher values for binomialblur - is it right?

binomialblur(5)

[Attachment 47260 - Click to enlarge]
binomialblur(50)

[Attachment 47261 - Click to enlarge]

And is it possible to clean purple stripe on right side?

[themaster1]

These are strong settings already. You can try to increase the value for ccd such as

CCD(30,1)

or

ttempsmooth (strength=8)

(that's the maximum), and

KNLMeansCL(h=8)

or higher

[gleim]

Yes, thank you. Increase value for KNLMeansCL gives effect. And what about the purple stripe on right side?

p.s.: I'm trying to encode with this script - fps is very low (about 0,5fps) and CPU Usage only about 10% - is it normal? CPU is Xeon E5-2670

[themaster1]

for the purple stripe on the side see this topic:
https://forum.videohelp.com/threads/385637-Erase-red-purple-mark-%28VHS-capture%29

[gleim]

themaster1, thank you. When discolored stripe is on the right side - that is work. But when stripe is on the left side - what should I do? I tried this:

# the width of the discolored right edge -- in Y pixels
EDGE_SIZE = 20
cropped = Crop(EDGE_SIZE,0,-0,-0)
extra = cropped.FlipHorizontal().Crop(EDGE_SIZE,0,-0,-0)
chroma = StackHorizontal(cropped, extra)
MergeChroma(last, chroma)

But in this case I'm getting an error: "MergeChroma: Images must have same width and height!"

[lordsmurf]

Standard chroma noise.
The Camcorder Color Denoise in VirtualDub should also be used for best results. Not just pure Avisynth.

[gleim]

lordsmurf
I use it already in my script:

LoadVirtualDubPlugin("C:\Program Files (x86)\virtualdubmod1.5\plugins\Camcorder_Color_Den oise_sse2.vdf", "CCD", 0)
CCD(10,1) # 0-100, default 30

But even maximal value (100) for CCD doesn't clean that discolor stripe.

That's why I'm asking how I should modify this script for the other side (left for example):

# the width of the discolored right edge -- in Y pixels
EDGE_SIZE = 20
cropped = Crop(0,0,-EDGE_SIZE,-0)
extra = cropped.FlipHorizontal().Crop(0,0,EDGE_SIZE,-0)
chroma = StackHorizontal(cropped, extra)
MergeChroma(last, chroma)

[themaster1]

Try this, it's the best i can do (rgbeq filter is included in this post)

AVISource("Pif_Sample.avi").assumetff()
## ADJUST Chroma to luma:
separatefields()
A=Last
B=A.Greyscale()
Overlay(B,A,X=3,Y=0,Mode="Chroma")
weave()

###1) Clean Chroma artefacts (heavy cleaning, spatio-temporal):
ConvertToYV16(interlaced=true)
orig=last
ev=orig.assumetff().separatefields().selecteven()
od=orig.assumetff().separatefields().selectodd()
ev
ue_chroma = UToY(ev).blur(0,1.5).ttempsmooth(maxr=6,lthresh=15 0, strength=6).KNLMeansCL(d=3,a=8,s=2,h=8)
ve_chroma = VToY(ev).blur(0,1.5).ttempsmooth(maxr=6,lthresh=15 0, strength=6).KNLMeansCL(d=3,a=8,s=2,h=8)
YToUV(ue_chroma, ve_chroma)
MergeLuma(ev)
ev_filtered=last
od
uo_chroma = UToY(od).blur(0,1.5).ttempsmooth(maxr=6,lthresh=15 0, strength=6).KNLMeansCL(d=3,a=8,s=2,h=8)
vo_chroma = VToY(od).blur(0,1.5).ttempsmooth(maxr=6,lthresh=15 0, strength=6).KNLMeansCL(d=3,a=8,s=2,h=8)
YToUV(uo_chroma, vo_chroma)
MergeLuma(od)
od_filtered=last
interleave(ev_filtered,od_filtered)
assumefieldbased().assumetff().weave()
###2) Clean Chroma artefacts (soft cleaning in rgb with CCD filter):
ConverttoRGB32(matrix="rec601",interlaced=true)
separatefields()
LoadVirtualDubPlugin("C:\Program Files (x86)\VirtualDub\plugins\Camcorder_Color_Denoise_sse2.vdf", "CCD", 0)
CCD(10,1) # 0-100, default 30
weave()
##############################################
###3) REMOVE COLOR STRIPE AT THE RIGHT (DESATURATE) :
X1 = 690
Y1 = 0
X2 = 720
Y2 = 576
#
X1 = X1 - (X1 % 4)
Y1 = Y1 - (y1 % 4)
X2 = (X2 + 3) / 4 * 4
Y2 = (Y2 + 3) / 4 * 4

Active = crop(X1,Y1,X2-X1,Y2-Y1)
Above = Y1 > 0 ? crop(0,0,width(),Y1) : NOP
Below = Y2 < height() ? crop(0,Y2,width(),height()-Y2) : NOP
Left = X1 > 0 ? crop(0,Y1,X1,Y2-Y1) : NOP
Right = X2 < width() ? crop(X2,Y1,width()-X2,Y2-Y1) : NOP
Last = Active
## PUT FILTERS HERE:
#
## Desaturate the PINK BAR on the right:
LoadVirtualDubPlugin("C:\Program Files (x86)\VirtualDub\plugins\rgbeq.vdf", "rgbeq", 0)
rgbeq("20G2020G2020G2020G2020G2020G2020G2020G0020G 2020G2020G2020G2020G2020G2020G2020G2020G2020G2020G 2020G2020G2020G2020G2020G1520G2020G2020G2020G2020G 2020G2020G2020G2020G2020G2020G2020G2020G2020G2020G 2020G2J20G2020G2020G2020G2020G2020G2020G2020G2020G 2020G2020G2020G2020G2020G2020G2020G1M20G2020G2020G 2020G2020G2020G2020G2020G2020G2020G2020G2020G2020G 2020G2020G2020G2020G2020G2020G2020G2020G2020G2020G 2020G2000G4100G4100G4100G4100G4100G4100G4100G4100G 4100H4H00H4H00G0J000a0")

IsClip(Left) ? StackHorizontal(Left, Last) : NOP
IsClip(Right) ? StackHorizontal(Last, Right) : NOP
IsClip(Above) ? StackVertical(Above, Last) : NOP
IsClip(Below) ? StackVertical(Last, Below) : NOP
return Last

converttoyv12(matrix="Rec601",interlaced=true)
## 4) remove some chroma bleeding:
mergechroma(awarpsharp2(depth=10, type=1, blur=1, chroma=4)) # chroma=4 ->luma guide

[LMotlow]

Try this, it works great with cartoons generally; require the filters: TTempSmooth, variable blur, KNLMeansCL, camcorder color denoise (vdub):

AVISource("Pif_Sample.avi")
assumetff()
###1) Clean Chroma artefacts (heavy cleaning):
ConvertToYV16(interlaced=true)
orig=last
ev=orig.assumetff().separatefields().selecteven()
od=orig.assumetff().separatefields().selectodd()
ev
ue_chroma = UToY(ev).blur(0,1.5).binomialblur(5).ttempsmooth(m axr=6,lthresh=150, strength=6).KNLMeansCL(d=3,a=8,s=2,h=6)
ve_chroma = VToY(ev).blur(0,1.5).binomialblur(5).ttempsmooth(m axr=6,lthresh=150, strength=6).KNLMeansCL(d=3,a=8,s=2,h=6)
YToUV(ue_chroma, ve_chroma)
MergeLuma(ev)
ev_filtered=last
od
uo_chroma = UToY(od).blur(0,1.5).binomialblur(5).ttempsmooth(m axr=6,lthresh=150, strength=6).KNLMeansCL(d=3,a=8,s=2,h=6)
vo_chroma = VToY(od).blur(0,1.5).binomialblur(5).ttempsmooth(m axr=6,lthresh=150, strength=6).KNLMeansCL(d=3,a=8,s=2,h=6)
YToUV(uo_chroma, vo_chroma)
MergeLuma(od)
od_filtered=last
interleave(ev_filtered,od_filtered)
assumefieldbased().assumetff().weave()
###2) Clean Chroma artefacts (soft cleaning in rgb with CCD filter):
ConverttoRGB32(matrix="rec601",interlaced=true)
separatefields()
LoadVirtualDubPlugin("C:\Program Files (x86)\virtualdubmod1.5\plugins\Camcorder_Color_Den oise_sse2.vdf", "CCD", 0)
CCD(10,1) # 0-100, default 30
weave()
converttoyv12(matrix="Rec601",interlaced=true)
## 3) remove some chroma bleeding:
mergechroma(awarpsharp2(depth=10, type=1, blur=1, chroma=4)) # chroma=4 ->luma guide

For more accuracy in colorspace conversion, you might want to know that the video sample is not interlaced. It's progressive film rate speeded up for PAL 25fps.