Update. Regarding 'Chroma_subsampling_notation.pdf', Figure 2, '4:2:2', I just tried to make a 2-pixel wide Cb0-1 that's centered coincident with the center of Y0. It can't be done. So Poynton must be assuming some sort of subpixel functions are applied without mentioning it. Certainly the quality of his presentation is lacking.
--Mark. On 16/07/2025 02.20, Yann Cainjo wrote: > Hi Mark > > Here are some article from Charles Poynton. > > https://www.poynton.ca/PDFs/Chroma_subsampling_notation.pdf > > https://www.poynton.ca/PDFs/Merging_RGB_and_422.pdf > > And his books... > > Best regards > Yann Hi Yann! Thanks for the links. That's in the spirit of friends helping friends. I like it! However, there are problems with what Poynton presents. Regarding 'Chroma_subsampling_notation.pdf', Poynton states, "The third digit has two possibilities. If the third digit is the same as the second digit, there is no vertical subsampling. If the third digit is zero, 2:1 vertical subsampling of both CB and CR is indicated." I ask, "What if the third digit isn't the same as the second digit and it isn't zero, either. What then?" Re: Figure 2, '4:2:2' and '4:1:1' and '4:2:0 (MPEG02 fr)' are shown centered on a luma pixel. That's impossible unless either 1: Subpixels are employed, or 2: Odd-even numerical averaging is employed. Subpixels are employed: Poynton says zero about subpixels. Certainly, no screen can show them without first creating a set of subpixels and then interpolating between them to derive the signal to the actual pixels, which can't move, but then, they're not the same Cb and Cr. If they're not the same Cb and Cr, he really needs to say it. Odd-even numerical averaging is employed: If this is true, then Poynton has some graphical deficiencies to clarify. Specifically, he needs to show not 4:4:4:4 but 4:4:4:4:4:4:4:4. Odd-even averaging is somewhat implied by discussions about aliasing, but again, he needs to say it. Certainly, odd-even scope averaging is the more probable method, except when it isn't. I will attempt to contact Charles Poynton to resolve these issues. --Mark. _______________________________________________ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email ffmpeg-user-requ...@ffmpeg.org with subject "unsubscribe".
