Vertical alignment with software instead of TBC?

[JELSTUDIO]

I have some old VHS-footage, captured without TBC.

The footage have a black bar on the left side.

I was thinking; could this be used to align the image, line by line, to reduce the vertical jitter somewhat?

I spent some time googling for ideas and software but couldn't find anything.

I believe it should be possible, using similar techniques as that used to fix rolling-shutter artifacts.
And especially since it doesn't have to analyze frames temporally to align them based on content (Which would be necessary had the black bar not been there), but can go 'simply' by making the contrast-area between the black-area and the image-area vertically straight (Of course some temporal adjustment might still be needed to avoid a possible flicker from side to side, depending on how the black-bar lines up over time or how/where you anchor the black-bar correction)

I know, theory is one thing, practicality another :)

But... could it be done?
Or even better; is there any software or any scripts for avisynth or Vdub that can already do this?

Thanks

[StainlessS]

JMac698 did a TBC thingy:- https://forum.doom9.org/showthread.php?p=1582177#post1582177

EDIT: Search on POSTS with search string "tbc" and username="jmac698" finds nothing, maybe search string too short or something.
Same result without username.
I had to use Google to find it, eg "tbc site:forum.doom9.org"

produces this lot:- https://www.google.com/search?ei=Q8POW7jULqnMgAa-moXoAg&q=tbc+site%3Aforum.doom9.org&oq=tbc+site%3Aforum.doom9.org&gs_l=psy-ab.3...58957.59421.0.59780.2.2.0.0.0.0.116.208.1j1.2.0....0...1c.1.64.psy-ab..0.0.0....0.C2aHpTf1rc0

EDIT: Google search seems to be case dependent, using "Site:" instead of "site:" brings up different results and not restricted to the forum.

[JELSTUDIO]

JMac698 did a TBC thingy

Thank you! That user-name led me on track to useful info :)

https://forum.doom9.org/showthread.php?t=162726&page=9

Sadly I can see now how the black-border method is not as useful as I thought it might be :( (Unless perhaps analyzed by some AI-software that 'sees' it more like we do, rather than by 'simply' detecting luminance-thresholds per line)

It seems like what's really needed is to analyze the entire image-frame, and temporally compare things also, to somehow 'restore' the correct frame-dimensions.

Maybe something comparable to a rolling-shutter-fixer or a frame-interpolator, but with some kind of image-detection 'intelligence' added (I don't know)