IO Data GV-USB2 + Panasonic ES15 sample

[boolian2]

Is it my eyes but those artifacts are not present in the later sample.

But something, as I suggested, in good day-light may provide better analysis.

Ok I will take another capture. In the meantime here is vdubbed version of the unprocessed capture in my last post. Notice there are artifacts in the "Wonka" letters

[Sharc]

I am probably way off the scale here but vdub2 reports the OP's UT captures as Rec 709. Surely they should be Rec 601 ?

Yes, the wrong codec was selected for SD video. Should be the 'UTVideo YUV422 BT.601' variant of the UTVideo codec suite, with the FOURCC code ULY2 instead of ULH2.
(This applies to the OP's previous captures as well, b.t.w.)

My bad. I changed it to ULY2. Here is a new capture using ULY2.

The capture in post #29 looks pretty good to me. In post processing you may want to reduce the saturation somewhat and increase the gamma slightly - a matter of personal preference though.

[boolian2]

New capture

[Sharc]

New capture

Looks good to me.
Still, you may want to check your cables to prevent any residual Y/C crosstalk (dotcrawl, rainbows) possibly caused by the cables (like in your previous anime video)

[boolian2]

New capture

Looks good to me.
Still, you may want to check your cables to prevent any residual Y/C crosstalk (dotcrawl, rainbows) possibly caused by the cables (like in your previous anime video)

Yeah I might replace the Monster Cable S-Video with a new S-Video cable from Amazon

[jagabo]

The latest cap is looking pretty good. For what it's worth here's a version with the worst of the halos reduced, edges smoothed, and a little temporal noise reduction.

LWLibavVideoSource("amarec(20220701-0305).avi", cache=false, format="YUY2") 
AssumeTFF()
TFM()
TDecimate()
ConvertToYV12()
FineDehalo(rx=3.0, ry=1.0)
FineDehalo(rx=6, ry=1.0, BrightStr=0.5, DarkStr=0.5)
Santiag()
Sharpen(0.3)
SMDegrain(tr=1, thSAD=200, refinemotion=true, contrasharp=false, PreFilter=4, mode=0, truemotion=true, plane=4, chroma=false)

[Alwyn]

Wow.

[jagabo]

A little better? Sharper and better aligned chroma, slightly sharper luma:

LWLibavVideoSource("amarec(20220701-0305).avi", cache=false, format="YUY2") 
AssumeTFF()
TFM()
TDecimate()
ConvertToYV12()
FineDehalo(rx=3.0, ry=1.0)
FineDehalo(rx=6, ry=1.0, BrightStr=0.5, DarkStr=0.5)
Santiag()
SMDegrain(tr=1, thSAD=200, refinemotion=true, contrasharp=false, PreFilter=4, mode=0, truemotion=true, plane=4, chroma=false)
MergeChroma(aWarpSharp2(depth=5), LanczosResize(width/2, height).aWarpSharp2(depth=15).LanczosResize(width,height))
ChromaShiftSP(x=-1)
Sharpen(0.3)

[Sharc]

@jagabo: Very nice.
Question about the chroma sharpening: Is the chroma subsampling (YV12, 4:2:0) the reason why you reduced the picture width by 2 before applying the sharpener?

[jagabo]

@jagabo: Very nice.
Question about the chroma sharpening: Is the chroma subsampling (YV12, 4:2:0) the reason why you reduced the picture width by 2 before applying the sharpener?

No. VHS just has very low chroma resolution horizontally. Reducing the width by a factor of 2 gets the horizontal chroma resolution closer to the vertical chroma resolution. That way aWarpSharp2 sharpens more evenly over the two axis. Otherwise a much higher depth value would be required and the "warping" would be too strong on the vertical axis -- with some smaller objects losing their color.

[Sharc]

@jagabo: Very nice.
Question about the chroma sharpening: Is the chroma subsampling (YV12, 4:2:0) the reason why you reduced the picture width by 2 before applying the sharpener?

No. VHS just has very low chroma resolution horizontally. Reducing the width by a factor of 2 gets the horizontal chroma resolution closer to the vertical chroma resolution. That way aWarpSharp2 sharpens more evenly over the two axis. Otherwise a much higher depth value would be required and the "warping" would be too strong on the vertical axis -- with some smaller objects losing their color.

Cool, thanks

[Sharc]

LWLibavVideoSource("amarec(20220629-2054).avi", format="YUY2") 
AssumeTFF()

n = TFM(clip2=QTGMC(FPSDivisor=2)) # normal field matching
s = TFM(Crop(0,0,-0,-1).AddBorders(0,1,0,0).AssumeBFF().SwapFields(), clip2=QTGMC(FPSDivisor=2)) # field matching after swapping field positions

ncombs = n.Crop(8,8,-8,-8).ConvertToYV12().FindCombing(0) # look for combing in n
scombs = s.Crop(8,8,-8,-8).ConvertToYV12().FindCombing(0) # look for combing in s
morecombs = Subtract(scombs, ncombs) # which has more combing?
ConditionalFilter(morecombs, s.Subtitle("s"), n.Subtitle("n"), "AverageLuma()", "LessThan", "126.0") # pick the frame with the least combing
TDecimate()

Where would I find the function "FindCombing()" which is called in above code? Thanks.

[jagabo]

It's something I wrote a while back. I think the usage of GeneralConvolution with YV12 video requires AviSynth+, not the older versions of AviSynth. Also since it uses mt_lut(), it could be modified to use mt_edge() with a custom matrix instead of GeneralConvolution().

###################################################
#
# Build a mask of areas where a pixel has a
# darker pixel above and below it.
#
##################################################

function FindCombing(clip v, int "threshold")
{
    threshold = default(threshold, 10)

    # find vertical low to high transitions
    Subtract(v, v.blur(1.0, 1.0))
    GeneralConvolution(0, "
        0 -8  0
        0  8  0
        0  0  0", chroma=false, alpha=false)
    lth = mt_lut("x 126 -")

    # find vertical high to low transitions
    Subtract(v, v.blur(1.0, 1.0))
    GeneralConvolution(0, "
        0  0  0
        0  8  0
        0 -8  0", chroma=false, alpha=false)
    htl = mt_lut("x 126 -")

    # Logical AND(ish)
    Overlay(lth, htl, mode="multiply")
    mt_binarize(threshold)
    mt_expand()
    mt_inpand()
    mt_inpand(chroma="-128")
}

##################################################

I also found that noise in some scenes of the later videos would sometimes cause ConditionalFilter() to choose the wrong frame. Lowering the threshold from 126.0 to 125.97 took care of that. You may have to tune it a bit for different videos.

But since your latest caps don't have the same field bounce problem you probably don't need the code.

[boolian2]

It's something I wrote a while back. I think the usage of GeneralConvolution with YV12 video requires AviSynth+, not the older versions of AviSynth. Also since it uses mt_lut(), it could be modified to use mt_edge() with a custom matrix instead of GeneralConvolution().

###################################################
#
# Build a mask of areas where a pixel has a
# darker pixel above and below it.
#
##################################################

function FindCombing(clip v, int "threshold")
{
    threshold = default(threshold, 10)

    # find vertical low to high transitions
    Subtract(v, v.blur(1.0, 1.0))
    GeneralConvolution(0, "
        0 -8  0
        0  8  0
        0  0  0", chroma=false, alpha=false)
    lth = mt_lut("x 126 -")

    # find vertical high to low transitions
    Subtract(v, v.blur(1.0, 1.0))
    GeneralConvolution(0, "
        0  0  0
        0  8  0
        0 -8  0", chroma=false, alpha=false)
    htl = mt_lut("x 126 -")

    # Logical AND(ish)
    Overlay(lth, htl, mode="multiply")
    mt_binarize(threshold)
    mt_expand()
    mt_inpand()
    mt_inpand(chroma="-128")
}

##################################################

I also found that noise in some scenes of the later videos would sometimes cause ConditionalFilter() to choose the wrong frame. Lowering the threshold from 126.0 to 125.97 took care of that. You may have to tune it a bit for different videos.

But since your latest caps don't have the same field bounce problem you probably don't need the code.

Which avisynth plugins are you using? I tried to find them but your code still wouldn't run. So I just tried Handbrake again, and this is how it turned out, does this look ok:

[Sharc]

Wrong Aspect Ratio and some blended frames.
If you seek help upload a snippet of the capture and post your script.

[boolian2]

Wrong Aspect Ratio and some blended frames.
If you seek help upload a snippet of the capture and post your script.

The filters are

Detelecine Default
Interlace Detection Default
Deinterlace Decomb EEDI2
Denoise NLMeans Medium Animation
Chroma Smooth Light Small

The Dimensions are

[Attachment 65719 - Click to enlarge]

[Sharc]

Can't help you with handbrake. I don't use it. Maybe someone else.
It's better you learn avisynth basics than wildly clicking options in GUI's
And upload a few seconds of the capture so someone may take a look.

[jagabo]

Which avisynth plugins are you using?

LSMASHSource, masktools2, TIVTC.

I tried to find them but your code still wouldn't run.

AviSynth usually gives you an error message indicating what went wrong.

So I just tried Handbrake again, and this is how it turned out, does this look ok

Your ITU cap has a pixel aspect ratio of 10:11. You encoded it as 1:1 so it displays incorrectly. Handbrake doesn't handle AVI with interlaced lossless YUV 4:2:2 chroma correctly. That has resulted in chroma blending artifacts. I don't think you can "fix" that about Handbrake.

[lollo]

It's better you learn avisynth basics than wildly clicking options in GUI's

+1

[boolian2]

Which avisynth plugins are you using?

LSMASHSource, masktools2, TIVTC.

I tried to find them but your code still wouldn't run.

AviSynth usually gives you an error message indicating what went wrong.

So I just tried Handbrake again, and this is how it turned out, does this look ok

Your ITU cap has a pixel aspect ratio of 10:11. You encoded it as 1:1 so it displays incorrectly. Handbrake doesn't handle AVI with interlaced lossless YUV 4:2:2 chroma correctly. That has resulted in chroma blending artifacts. I don't think you can "fix" that about Handbrake.

Do you know which filter in Handbrake is causing the artifacts? I can just try turning it off until I figure out the avisynth stuff

[davexnet]

Try turning off the interlaced options, just leave detelecine

[boolian2]

Try turning off the interlaced options, just leave detelecine

I tried just detelecine first, without deinterlace, and it looks like Handbrake's "detelecine" isn't "inverse telecine", it just cleans up artifacts it finds from the original telecine process when the movie was put on VHS. So the "detelecine" didn't deinterlace it

[davexnet]

Did you set the frame rate to 23.976? Otherwise post a sample, something with
steady movement

[boolian2]

Did you set the frame rate to 23.976? Otherwise post a sample, something with
steady movement

Yeah, the frame rate is set to 23.976. I turned off Chroma Smoothing and Denoise and the artifacts are still there:


[Attachment 65727 - Click to enlarge]

Then I changed the deinterlacing from EEDI2 to Bob and it came out with this which I don't know how to explain:

[jagabo]

Do you know which filter in Handbrake is causing the artifacts? I can just try turning it off until I figure out the avisynth stuff

None of the filters you can control. It's the initial conversion from YUY2 to YV12. It's treated as progressive instead of interlaced, resulting in the chroma blending.

[jagabo]

Then I changed the deinterlacing from EEDI2 to Bob and it came out with this which I don't know how to explain:

The field order is wrong. It was treated as bottom field first rather than top field first. And the chroma blending is still there.

[Sharc]

Wrong field order. Just continuing with wildly clicking around in options of handbrake or other GUIs doesn't lead you anywhere.

[boolian2]

Do you know which filter in Handbrake is causing the artifacts? I can just try turning it off until I figure out the avisynth stuff

None of the filters you can control. It's the initial conversion from YUY2 to YV12. It's treated as progressive instead of interlaced, resulting in the chroma blending.

I downloaded the avisynth plugins you're using, what is _xp? I installed the plugins and got this error message when trying one of the avisynth scripts you posted earlier

[Attachment 65729 - Click to enlarge]

[Sharc]

http://avisynth.nl/index.php/FineDehalo
But it won't solve the basic problem with your source or script. Means you can leave it away for the time being.

[boolian2]

http://avisynth.nl/index.php/FineDehalo
But it won't solve the basic problem with your source or script.

So that's not included in the plugin? Do I just copy and paste it at the top of the file?