Quote:
Originally Posted by Geoff D
Good job Sony's higher-end sets have 14-bit processing then to actually deliver true 10-bit performance, which is another thing that's been holding back the precision of certain displays and display types. You can have all the near-black shades you want but if the processing's bit depth is choking on it, well....
|
Yes.
According to my little table, and the way Charles Poynton calculated linear,
12bit 10k PQ = 13bit log = 15bit gamma = 24.3bit linear
10bit 10k PQ = 11bit log = 13bit gamma = 21bit linear. If you discard the top, 2000 nits 18bits linear, 1000 nits 17bits linear.
Now, the DAC stage, so the electrical signal drives the panel, is it linear or there is some kind of stage that massages the signal to fit the EOTF curves of the LCD/OLED/DLP, saving bits?
In a CRT the video signal, as it was inverse gamma encoded, went -> DAC straight to the gamma curve of the tube, and the tube painted the picture, so no digital values transposing (and truncating) from one curve to another.
theetofofstuff.jpg
A LCD curve is different to CRT gamma; an OLED seems more similar in shape to a CRT in that composite, but with a higher slope (or "gamma"), which from this seems something like around a gamma of "3.2", which probably is a better fit for PQ as PQ has a much higher slope than γ2.4. But since none of these EOTF curves are shaped like the inverted
S of the PQ one, somewhere between the video digital codes and the panel physical response, the digital values of the source levels and/or the shape of the electrical signal have to be recalculated/manipulated, and if in the digital domain you need the extra bits to not truncate the perceptual coding advantage of PQ, unless you had a panel whose EOTF mimicked PQ directly.