On Wed, Jan 8, 2020 at 06:07 Roger Breton via colorsync-users < colorsync-users@lists.apple.com> wrote:
Wire,
I'm not really after one "best" visualization. MacAdam ellipses are daunting to me, I won't go there unless I have to. I realize no color space will ever have perfect qualities but that's not what I'm after... My point, in raising the question with Steve yesterday, was what kind of numerical abstraction can be applied to distinguish between "unique" colors -- sorry to sound like a broken record in the context of color management? [...]
The 16 million colors thing is overloaded. In the context of your queries, all it is about is a data format for encoding of stimuli (the light coming out of your display) that ensures sufficient variety to produce smooth gradations. Because we are talking about the display's data interface, and because you can count the stimuli produced for a pixel input value 1-to-1, and because as stimuli — using this word advisedly — the effects are totally within a visible range, the count of a pixel value at the display input is considered to be a count of colors the display can produce. That's it. No further parsing or interpretation is needed. You don't want all the inputs to be perceptually distinct, because you want gradients to be smooth, but for cost you don't want way more than you need for sufficient response. In the early days of graphics, pixels and colors were scarce and expensive and color was coarse. Now its possible to over-provision values, and new ranges of brighter higher gamut output are available. Color is rich and smooth. Given the traditional limits of the device data formats, there are other ways of getting the most out of limited pixel color values. For example, you can vary the tonal response curve, aka gamma, and you can use pixels in groups: dither and subpixel rendering are common tactics to get perceptually smoother overall response from a system at limits of visible quantization artifacts confess, initially, I had this 16.7 million RGB "color combinations"
question in mind. But, like many here, I am aware of research that concluded that men and women with "normal color vision" can only discriminate 10 million different colors (or something like that?) which is referred as the "Gamut of real colors", I think, see Dr Michael Pointer -- a nice guy, btw 😉 I never read the papers about this concept (I have no idea how that number was ever arrived at or very vaguely) but I took it "on faith" that this estimate made intuitive sense to me.
There are only 24 hours in a day.
/ Roger
I find the idea of countable colors in sense of max perceivable to be at edge of a red herring. Ask why you want an answer to this question... and if you don't know exactly why you might abandon it for now and see where it pops up again. Ya I'm interested in idea of Pointer's gamut too but I don't think it involves counting.