On 20/05/2025 15.37, Erik Dobberkau wrote:
I wrote:
The CinGG CLUTs do everything in one step, so, no, IIRC you don't need to
tonemap because they tonemap. They take input yuv pixels from the input
colorspace and output yuv pixels in the output colorspace.
IIRC, the missing piece to map from 2084 PQ to SDR may be that you need to
know are the maxFALL and maxCLL values to have a frame of reference (your
virtual 1.0 clipping point), because the content has rarely been color
graded for the theoretical 10,000nt display...
maxCLL and maxCLL are from libplacebo. I don't know what they are, but it occurs to me that for
standard stuff like what's found on UHR BD and HR BD they must be known constants and therefore
already built-in.
Doesn't anyone normalize nits? -- 100 nits, 1000 nits, 10000 nits, they're just scaling factors,
right? I've attached 'PQ v HLG v gamma.png'. I normalized nits and I put the right-hand graph in
it's correct orientation. I've seen other HLG transfer graphs. They were not normalized and they had
input on the vertical axis and output on the horizontal axis. Nothing like flipping axes to keep the
novices confused, eh?
_______________________________________________
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".
