I learned in elementary school that you could get green by mixing blue with yellow.

However with LEDs, TFTs, etc. you always have RGB (red, green, blue) values?

Why is that? From what you learned in elementary yellow would be the 'natural' choice instead of green.

  • 3
    $\begingroup$ You should rather ask why they keep teaching this nonsense in elementary school. It's only right when calling magenta red and cyan blue, which just aren't the same things. $\endgroup$ Sep 18 '11 at 15:16
  • $\begingroup$ @leftaroundabout Printers' "cyan" and "magenta" have as much right to be called blue and red as sRGB's "blue" and "red" primaries. "Cyan", especially, is much closer to unique blue (≈sky blue) than sRGB "blue", which is actually quite purplish and is probably what Newton called indigo. $\endgroup$
    – benrg
    Jul 31 '16 at 22:32
  • $\begingroup$ @benrg I'm pretty sure the deep shades of blue such as ultramarine and prussian were always considered as blue, in its most intense form. It only makes sense to anchor that meaning and give the less saturated blueish hues individual names. And even if all this is historically not so clear-cut: today the terminology of RGB and CMYK is firmly established and can well be taken as a defining system for the colours themselves. $\endgroup$ Jul 31 '16 at 23:49

When you mix colors using Watercolors, then they mix as "Subtractive Colors". However, Light itself mixes as "Additive Colors".

Even though it might seem strange why the inherently same thing works so differently, it makes sense if you think about Watercolors, etc. as absorbing everything but that specific color.


What @Michael said, plus your retina has sensors for roughly red, roughly green, and roughly blue, and not for any other colors. (BTW, the green sensors are more sensitive, so it takes less green to make the same brightness.)

When you see something yellow, it's in between red and green, so it excites both of them, and your visual cortex happens to call that combination yellow.

If your TV set turns on a red and a green pixel so close together that you can't tell they are separate, what does your brain call it? Yellow, because there's no way it can tell the difference between real yellow and red plus green.

Paints work by subtracting colors, not by adding them. Blue subtracts red and green, and yellow subtracts blue. When you mix blue and yellow paint, the color that is least-subtracted is green. Colors you get by mixing paints tend to be muddy, as opposed to brilliant.

  • $\begingroup$ But, what if you actually see yellow light, for example if you look into a spectroscope, you'd see all the visible colors. Does true yellow stimulate both the red and green sensor? $\endgroup$ Apr 23 '12 at 12:45
  • $\begingroup$ @50ndr33: Yes it does, because the red sensor responds to a broad distribution of wavelengths, heaviest in the red. Green, same way. Check this out. $\endgroup$ Apr 23 '12 at 17:22
  • $\begingroup$ "your retina has sensors for roughly red, roughly green, and roughly blue, and not for any other colors" is wrong. The three cone types have broad, overlapping sensitivity curves that between them cover the entire visible spectrum. It is not at all true that they sense red, green, and blue specifically. $\endgroup$
    – benrg
    Jul 31 '16 at 18:40

The inherent difference is things that emit colors, e.g. LEDs, and things that place substances on a surfaces to color them, e.g. printers. In the latter case, the CMYK (cyan, magenta, yellow, black) space is commonly used, rather than RGB, so you were on the right track with yellow. BTW, black is there because the black produced by mixing C, M, and Y doesn't look as good and printers generally print black stuff on white paper).

On a philosophical note: color is not a physical property of an object, i.e. we should not really talk about a "blue car"; it is an example of conscious experiences known as qualia.

  • $\begingroup$ ""color is not a physical property of an object,"" Strange, how do You explain the reproducible measurement of body colors? $\endgroup$
    – Georg
    Sep 19 '11 at 10:27
  • 1
    $\begingroup$ @Georg, spectrum reflected by an object is a physical property. What "color name" we perceive that spectrum as, is not. See blog.xkcd.com/2010/05/03/color-survey-results for an interesting informal survey. $\endgroup$
    – ptomato
    Sep 19 '11 at 10:38
  • $\begingroup$ The "color name" is as reproducible as the physical measurement is. (if You are not color blind) You learned to call a body "red" from Your elders. The problem is, that nobody knows whether the thing You see in brain is the same as others peoples impression. But that does not mean that bodys do not have a physical property named color. $\endgroup$
    – Georg
    Sep 19 '11 at 10:51

Because these 3 colors are the only ones that the human eye can see. We have 3 kinds of cone cells in our retina, each one reacts to a specific wavelength.


As you can see in this article, it happens that these three wavelengths correspond approximately to red, green and blue.

Since our eye cannot see more than 3 wavelengths, you don't add a lot of information if you add a fourth color. 3 seems like a good tradeoff between space needed to store a picture and fidelity.

I our eyes had more types of cones (like cones sensitive to infrared or ultra violet) we would certainly use more colors to encode our images.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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