Sorry, I wasn't trying to question your qualifications.
Here's what you said to me above, in reference to the Wikipedia link you posted:
oh boy, actual color science! not the fake "color science" term everyone on dpreview likes to misuse.
Right, as in "oh boy I get to discuss color science with an expert" not "the fake color science that's all over this forum.
It wasn't directed at you, sorry if I offended you.
edit: The "fake color science" I am referring to is the term people use all over DPReview (not this specific forum, that's why I said DPReview) to say why they like the output of one camera brand over another.
OK, sorry. I thought the antecedent of actual color science was the WIkipedia link you posted.
Let me take my best shot at answering a question that is similar to the one you're asking. It may even be the question you're asking, but I can't tell for sure.
For the purpose of enumerating colors, let us say that two colors are the same if a normal human can't tell the difference between them, and different if a normal human can. Let us further say that the color spaces under discussion here all have nonlinear tone curves. Those tone curves might be the same from one space to the next, or they might be different. Let us further say that we're talking about RGB additive color monitors, with physical primaries.
If we discuss humans observing the real world, humans can discriminate among about 10 million colors. Translating this to RGB monitors is tricky, because no commercial RGB monitor, no matter the precision, can display all the colors that people can see.
As the gamut of RGB monitors increases, more and more visible colors can be seen, but with modern monitors, the range of output spectra producible by the monitor exceeds the number of countable colors. For example, sRGB with 8-bit precision can produce almost 17 million different spectral outputs. However, not all of those spectral outputs are perceivable as different colors. Banding is hardly ever a problem with 8-bit sRGB. That's a big part of the basis for choosing it as a least common denominator.
However, as the primaries of the representation grow further apart, it takes greater precision and/or different tone curves to represent colors with no banding. For example, with PPRGB, 8 bit precision often results in banding, and the standard for precision in photo editing of PPRGB images is 15 bits plus one state. Same with CIEL*a*b*. Both of those have gamuts in excess of any commercial monitor. In fact, PPRGB can encode values not recognizable by humans at all.
The peak brightness of the display and the tone curve of the encoding and of the display also affect the precision necessary to avoid banding. If the black point remains constant, a brighter display will show more banding than a darker one.
So the answer to the 8-bit vs 10-bit precision question and the number of colors displayable is complicated.
With wide-gamut monitors with high brightness and low black points, two things need to change from what we're now calling SDR in order to avoid banding. First, the precision needs to increase. But going from 8-bit to 10-bit precision is not sufficient with gamuts like Rec 2020 at brightnesses on the order of 4000 cd/m^2. The tone curves must also be modified, which is what the HDR standards that I know about do.
Does that help?
Jim
.
