Help with odd Jitter baked into source laserdisc, Domesday doesn't fix it

[aramkolt]

Hello all, have a kind of odd laserdisc jitter during capture that I've never encountered before and I am wondering what can be done about it. I am fairly certain that it is burned into the source release of this film, see below for more information.

The background is that this is the Japanese 1984 laserdisc release of Urusei Yatsura 2 - Beautiful Dreamer. There were some people a lot smarter than me that wanted to restore the film back to it's original 4:3 aspect ratio and apparently the later releases to include other laserdisc releases and both DVD releases had some sort of odd frame blending and color issues. The BluRay release does not have any jittering or issues (and looks amazing), but they cropped it to 16:9 ratio which essentially discards something like 30% of the original image.

Anyhow, the group that tried to restore this did captures with a DomesdayDuplicator of 3 different discs and averaged them all together to account for any laser-rotted areas which they all seemed to have. The way laserdiscs are made, they are like vinyl records and the "pressings" are identical to each other as they are formed from the same mold. The result was this same odd jitter than I'm getting via a traditional composite capture, though for whatever reason, my copy does it slightly less. Perhaps this wasn't really visible in the days of CRT when this would have been normally watched, but it's pretty obvious on modern screens.

Luckily, my copy does not seem to have significant laser rot, but I'd like to see if I can do any better with the jitter aspect of it. It could also probably use some de-ringing. The film should be inverse telecined as far as deinterlacing goes, but I've provided a sample that is NTSC interlaced ProRes422HQ from an SDI capture chain.

Any suggestions on how to fix this type of jitter or sample scripts would be greatly appreciated!

[Selur]

vs-temporalfix (with ludicrous high values) seems to help a bit.
Just throwing random stuff at it,.. (this doesn't seem like a real solution, since encoding those 451 input frames took ~100min. )

Cu Selur

[jagabo]

This reduces the jitter by quite a lot:

LWLibavVideoSource("JitterSample.mov") 
AssumeBFF()
TFM(pp=0)
TDecimate()
ConvertBits(8)

e = SelectEven().Stab().QTGMC(InputType=2)
o = SelectOdd().Stab().QTGMC(InputType=2)
Interleave(e,o)
QTGMC(InputType=2)

[Selur]

Nice!

[jagabo]

Adding SourceMatch=3 and Lossless=2 to all the QTGMC calls jitters a little more but keeps a little more detail. One might want to turn down the sharpness setting too.

[Selur]

using 'TFM(pp=0)' and later QTGMC for cleaning is the main thing

[aramkolt]

That really is amazing! I'll share the script with the project creators and see how it works for them!

Is there a particular script for de-ringing that would be optimized for the sort of ringing you can see on the dark lines on white background? I understand that de-ringing often causes a lot of detail loss, but wondering what is suggested as a happy medium?

Thanks again!