reference display luminance characteristics

We already have the mastering display luminance characteristics but we should also have the reference display luminance characteristics.

See also: pq/color-and-hdr#23

!44 (closed) is an attempt to solve this. The w3c is currently also trying to figure this one out, so leaving this open.

mentioned in issue #13 (closed)

mentioned in issue #12

mentioned in issue #30

!74 (merged) is the new take.

I've been pondering what does the min luminance mean. It might be backlight leaking, or it might be flare. Essentially, at pixel code point zero we have some non-zero luminance, a baseline. When some pixel color channel value goes one step above zero, surely the other channels keep on leaking like they used to. The non-zero color channel just adds a bit more light at the chromaticity of the respective primary. Therefore min luminance is (hopefully) whitish baseline light emission, meaning that dark shades cannot reach the saturation of the primaries even strictly colorimetrically. This means the chromaticity of the stimulus is not the same as... the pixel's encoded chromaticity?

I believe a lot of monitors have a setting for the black level because of that. This could also done at a compositor.

I'm thinking of the disconnect between the pixel value encoded colorimetry, and the actual stimulus colorimetry on the reference display of the encoding.

We can decode pixel value encoded colorimetry, assuming min luminance is literally zero, and arrive at some CIE 1931 XYZ. Then we can add the effect of min luminance by adding the XYZ of the min luminance to that. This should give us the intended stimulus in the encoding reference environment. (Sometimes the two steps are not separable, like in BT.1886 with non-zero L_B.)

Then the reverse of that when re-encoding for another display.

I wonder if that's actually done in the industry, and should we do that. I've seen a few papers set L_B=0. We also may not be aware of the right L_B to use for a destination display.

Mh, I think the encoded colorimetry describes the intended stimulus in the encoding reference environment?

So leakage and flare (i.e. min luminance) included? Perhaps.

OTOH, if you profile a monitor, don't the colorimeters usually block flare? So min luminance is only leakage. Depends on the instrument type I guess.

Mh, yeah, leakage is definitely included in the measurements. I guess the viewing environment might be chosen such that the leakage overpowers flare? Or just use an instrument that also catches flare to some degree? Lots of speculation and I'm not sure where to find better information.

A thorough read of https://www.itu.int/dms_pubrec/itu-r/rec/bt/R-REC-BT.814-4-201807-I!!PDF-E.pdf might help.

Read it. Was interesting. Didn't answer the questions I had.

mentioned in merge request !99 (merged)