On Fri, Jan 3, 2025 at 12:50 PM Seeger, Chris (NBCUniversal) < Chris.Seeger@nbcuni.com> wrote: > I vote for releasing cICP first and delay mDCV and cLLI. > It's trivial for me to implement those chunks and both are pretty much essential for producing correct test files, cLLI for PQ encoding with sRGB data (which would come out very dark without cLLI since the PQ 1.0 corresponds to 2000cd/m2; this is assuming I understand H273). mDCV when data with a different white point is encoded in a colour space, particular colour space 10 (SMPTE ST 428-1). Without those chunks the display will be confusing in a conformant decoder; bad for testing. I also just pointed out to ChrisB that the libpng release and development branches are currently broken and write cICP in the wrong place then, under certain circumstances, reject it on read. Beware of untested software :-) John Bowler >
Received on Friday, 3 January 2025 20:59:46 UTC