It's a common conception that there are at least three different colors which produce most of the visible spectrum.

I am wondering if there is a unique set of these "primary" colors which could be chosen as the best one.

For instance, what's the difference between the sets {Red, Green, Blue} and {Red, Yellow, Blue} and if I were to choose a better option, which one would I pick?

Series of comments pointed out that the question is not so clear.

Firstly, let's say that we're calling the human-visible spectrum (which is, more or less, defined (yet, empirically) between certain wavelengths) a color.

Now, using terms red, green, blue, yellow, etc. I mean the wavelengths (rather, localized wave packets around some wavelengths) commonly associated with these names. The names are not of any importance, really.

Addition of several combinations of localized wave packets (i.e. colors) are possible in different situations.

By mixing colors I mean the event when a human mind cannot differentiate the superposition of the several wave packets from a single localized wave packet (lying at a, generally, different wavelength in the spectrum).

For instance, the perception of the superposition of $\lambda_1$, $\lambda_2$, $\lambda_3$,... wave packets (with known intensities, of course) can result into the same brain signals which are resulted from a single $\lambda$ wave packet (again, with some intensity).

  1. Is this $\lambda$ unique? Meaning that for each human (excluding, maybe the colorblind ones) a predefined set of wavelengths and intensities would result to the same wavelength/intensity of the "mixed" color?

  2. If I want to construct the most number of different wavelengths for these "mixed" colors (taking into account that the set is continuous, I'd talk about the greatest measure for the subsets of the visible color spectrum rather than "most number" of colors), a) what is the least number of finite wavelengths which could, by combining them with different intensities, obtain such a result? b) where do these finite wavelengths, most likely, reside in the visible spectrum?

    (For instance, if you'd answer 3 colors: RGB, I would like to know why and if they can produce more colors than the RYB or some random choice of 3 different colors.)

  • 1
    $\begingroup$ How do we know that we all see the same hue, due to the cellular structures, receptors, corneas etc of our eyes, it's seems inevitable to me that my green will be slightly different than your green . $\endgroup$
    – user108787
    Commented Jul 11, 2016 at 21:10
  • $\begingroup$ en.wikipedia.org/wiki/Primary_color $\endgroup$
    – user65081
    Commented Jul 11, 2016 at 21:13
  • $\begingroup$ I'm not talking about a color as a human-perceived phenomenon. I'd consider colors and their combinations as wave packets with some energies, etc. Regardless of what we might call green or blue or whatever, (if my understanding is correct) some combination of wave packets is perceived as if it were a single wave packet (single color). $\endgroup$ Commented Jul 11, 2016 at 21:14
  • 3
    $\begingroup$ If you are not talking about human perception, then I don't know what you're asking about. $\endgroup$
    – ACuriousMind
    Commented Jul 11, 2016 at 21:22
  • $\begingroup$ There is no color in the physical world, it is only on the eye of the beholder $\endgroup$
    – user65081
    Commented Jul 11, 2016 at 21:22

2 Answers 2


The light spectrum is continuous, RGB is about how you perceive light: You have 3 detectors in your eyes, each one is sensitive to a different range: one in the short wave lengths (Blue), one in the long (Red) and one in the middle (Green). When yellow or orange or any other light gets to your retina, each one of the detectors sends a signal and by the comparative strength of the signal from the red, green and blue detectors your mind knows which color came into your eye.

Again, there is yellow light, orange light, green light, violet light and everything from the electromagnetic spectrum, the RGB is just how your mind perceives colors.

Here is a graph that I found in Google that shows the sensitivity of the detectors: http://www.normankoren.com/Human_spectral_sensitivity_small.jpg


At a physics level, there is no way to "mix" two colors to get another. At a physics level, a color is typically defined by the intensity of the light at each frequency. It is a continuous function, which cannot be replaced. This continuous distribution can be shown by passing the light through a prism, which changes the angle of the light in a frequency dependent way:

enter image description here

If you attempt to create white light using three "primary" colors, you find the resulting distribution is very different:

enter image description here

Note the lack of continuity between the colors. This light is very different than the white light above. However, to us humans, the colors look the same. This is because we perceive colors using 3 (or 4) different chemically sensitive compounds and then process it in the brain from there. To a mantis shrimp, which processes light very differently and has 12 different chemicals, our "white light" LEDS would look very not-white.

When it comes to humans, we have found a particular model of color valuable, known as CIE. The goal of CIE was to make it so that every color a human "distinguishes" has about the same area on the graph. The CIE graph looks like:

enter image description here

As it turns out, if you mix two colors on this graph, the perceived color is a simple average of the colors you mixed. Thus, if you have 3 pigments to play with, you can generate any color within that triangle. Everything outside of that region is "out of gamut."

From this perspective, for humans, red green and blue create a triangle that covers the most area, so thus can emulate the most colors. However, this only true for human perception. At the physics level, colors just don't mix that way.

  • $\begingroup$ Also, the three sensor color bands overlap,. $\endgroup$ Commented Jul 11, 2016 at 21:48

Not the answer you're looking for? Browse other questions tagged or ask your own question.