0

Human eyes have retinal cells that are sensitive to red, green, and blue light. Pixels on electronic screens use the same three colors to produce many apparent colors.

But a little thought (and knowledge of vector spaces) reveals that neither of these facts necessitates the other. There are infinitely many bases for a single vector space, and there are infinitely many sets of primary colors that could reproduce a large color gamut.

So why was RGB chosen as the basis / primary color set for pixels? Was it technologically easiest to produce pixels that create those colors? Is there a practical advantage to having the colors align closely with those that our retinal cells can detect? Or is there an external factor that singles out RGB as a "preferable" color basis for both vision and pixels (e.g., perhaps RGB allows for the largest color gamut).

Similar question here, but with unsatisfying answers. Very related to question #3 here, but it isn't addressed directly in the answers from what I can tell. That question is about why the frequencies of pixels are slightly different from those that our retinal cells are most receptive to, which adds yet another wrinkle to this topic.

BioPhysicist
  • 56,248
WillG
  • 3,315
  • All of the questions you ask in your third paragraph seem more suited for engineering than for physics (there are many users who will still answer though; I could probably predict who as well). – BioPhysicist Jun 07 '21 at 16:44
  • @BioPhysicist It depends on what the answer turns out to be. Some sources I've seen (see 2nd link) suggest that there are fundamental (physics) based reasons for preferring an RGB basis. – WillG Jun 07 '21 at 16:48
  • I wasn't saying there is no relevance to physics here. I am saying the question you have decided to focus on are more geared towards engineering. What fundamental physics are you specifically asking about here? – BioPhysicist Jun 07 '21 at 17:00
  • I was getting at the connection between the choice of primary colors and the size of the color gamut. Seems like that's what BowlOfRed's answer here is focusing on too. – WillG Jun 07 '21 at 17:09

1 Answers1

4

The CIE Chromaticity diagram from that other answer is somewhat informative.

enter image description here

With 3 display colors, you can perform a linear interpolation of them, and you can reach any space within the convex hull of their locations. In other words, if you pick 3 points, you can draw a triangle between them and reach any color in the interior.

There are certainly material reasons that make one specific color easier or more difficult to produce, but in general you can see that the largest triangle you could create within the CIE color space is one with one corner at red, one corner at blue, and one corner in green.

You can cover more space with more points, but that becomes more complex with less marginal benefit. 3 points seems to be a good tradeoff between utility and technical complexity.

See also this SO question enter link description here which shows a representation of a few different color gamuts.

enter image description here

BowlOfRed
  • 40,008