mandarinka
14th February 2012, 00:11
I have been confronted with this video defect. I wasn't able to identify or solve the problem at all, so I'd like to humbly ask for some help...
Shot 1 (http://i.imgur.com/LpXtN.png)
Shot 2 (http://i.imgur.com/9d9qf.png)
Shot 3 (http://i.imgur.com/9zYAU.png)
Shot 4 (http://i.imgur.com/Jw5h4.png)
It looks like chromatic aberration, but the distortion doesn't extend radially from the centre, rather it seems to me that the strength increases horizontally from right (where there's no colour mismatch) towards left side of the frame (where chromatic aberration-like effect is clearly visible). I guess it's best described as red and green halos on the left and on the right of an edge.
I tried to resize chroma planes to see if this can fix the mismatch, but it doesn't seem that can help (it doesn't seem the whole chroma plane is misaligned like that). The colours of the halos also seem to suggest that converting to rgb and doing something like what fixchromaticaberration() won't help either (I tried it and the the colour plane realigning it does doesn't match the misalignment in the source).
Can anybody tell me what exactly is going on here (is it really chromatic aberration effect, or is the haloing caused some other way), and more importantly, is there any hope for fixing those halos?
The source is a dvd release of an OVA, but it's from the first half of 1990s, so the master probably came from an earlier laserdisc release.
The pictures I posted above are taken from my work-in-progress filtered encode (crop, bifrost/tcomb against dot crawl, deinterlacing with qtgmc which further supresses dotcrawl and rainbows and finally srestore) so that the haloing stands out more, but here are samples from the original dvd stream matching those screenshots: zip at mediafire (8MB) (http://www.mediafire.com/?n9gd20czhtwd93w)
Shot 1 (http://i.imgur.com/LpXtN.png)
Shot 2 (http://i.imgur.com/9d9qf.png)
Shot 3 (http://i.imgur.com/9zYAU.png)
Shot 4 (http://i.imgur.com/Jw5h4.png)
It looks like chromatic aberration, but the distortion doesn't extend radially from the centre, rather it seems to me that the strength increases horizontally from right (where there's no colour mismatch) towards left side of the frame (where chromatic aberration-like effect is clearly visible). I guess it's best described as red and green halos on the left and on the right of an edge.
I tried to resize chroma planes to see if this can fix the mismatch, but it doesn't seem that can help (it doesn't seem the whole chroma plane is misaligned like that). The colours of the halos also seem to suggest that converting to rgb and doing something like what fixchromaticaberration() won't help either (I tried it and the the colour plane realigning it does doesn't match the misalignment in the source).
Can anybody tell me what exactly is going on here (is it really chromatic aberration effect, or is the haloing caused some other way), and more importantly, is there any hope for fixing those halos?
The source is a dvd release of an OVA, but it's from the first half of 1990s, so the master probably came from an earlier laserdisc release.
The pictures I posted above are taken from my work-in-progress filtered encode (crop, bifrost/tcomb against dot crawl, deinterlacing with qtgmc which further supresses dotcrawl and rainbows and finally srestore) so that the haloing stands out more, but here are samples from the original dvd stream matching those screenshots: zip at mediafire (8MB) (http://www.mediafire.com/?n9gd20czhtwd93w)