Valerii Zapodovnikov
Valerii Zapodovnikov
> there are new inputs I think it is already good. Just ping the devs in ffmpeg. And this one is NOT actually breacking ABI.
@remia FYI. I patched ffmpeg's mxf_jp2k_rsiz https://github.com/FFmpeg/FFmpeg/commit/91ba771a3fee9757273416789ba01e1a7c62890b Now, I suppose that may be not enough, but whatever. Will test and update this comment. Yep, does not work with ffmpeg -v...
Relanded in https://github.com/uclouvain/openjpeg/pull/1236. So the issue is "fixed". Really, though, @acdha, this is bad. Adobe RGB (1998) for one Y component... Hmm.... Theoretically it can be done. The profile looks...
> no argument that the file is malformed I do not agree... I mean sGray 2.2 gamma uses sRGB white point and black point but Adobe RGB has a different...
Was not it removed recently?
Does not happen on acrobat for android so an obvious bug.
Was the sample crafted? Will extract jp2 file from the pdf. 
There is also a problem that sYCC decoding is not bit perfect between ffmpeg's libopenjpeg wrapper and ffmpeg native decoder jpeg2000. https://drive.google.com/file/d/1A4k7p2nYMom-ytksrnPbs-JUfsB74keD/view?usp=sharing You can try to use ffmpeg -vf extractplanes=y...
> will just convert luma without bllue and red (it There is no blue and red. Also all grayscale is ONLY Y component.
> either this test case needs refreshing You need to check whether those pixels that are wrong are out-of-gamut for BT.601 full range matrix (and that is actually a lot...