When I was a kid in art class, they taught us that the primary colors were red, blue, and yellow, and that these were so-called because combinations of them could produce any color.
Maybe this is a stupid question with an obvious answer, but why (and how) then, do TVs and computer monitors represent all the colors with red, green, and blue. How do they make yellow?
While reading a little bit about digital video on the web, I came upon a transcribed lecture from a professor at a major university (can’t remember which one) who said that RGB were the primary colors. Now I’m really confused.
You are absolutely correct in your assertion that red, blue and yellow are primary - IN SUBTRACTIVE color. That is, the pigments ABSORB light at various frequencies. On your monitor (or in the spectrum), you’re dealing with ADDITIVE color. The three INKS make black. RGB light together makes white.
A finer note: The subtractive colors are magenta, cyan and yellow. Your “red” is actually a bright pink, your “blue” is a very light shade of blue, and “yellow” is very bright yellow. Added together they make black.
Shift your color wheel one click to the right to get RGB: Your red is bright red (magenta plus yellow), your blue is deep blue (cyan plus magenta) and your green is vivid green (cyan plus yellow).
Well, shit. Bibliophage and Walloon scooped me whilst I was crafting my reply, and the master explained it all.
Motto to self: Type faster or go to bed.
D’oh! I can’t believe I asked a question Cecil already answered. Thanks for the info and links guys, and sorry for not searching the archives before I posted.
Well, I was about to post the experiment using a drill, but I see Cecil’s already covered that. So what I will add is that the reason you have black in the CMYK printing process is because, well, printing isn’t perfect and when you mix equal proportions of CMY you’re more likely to get some awful shade of brown rather than black. Also, CMYK doesn’t reproduce all colors well. For example, orange (as I’ve discovered) is notoriously difficult to reproduce with the CMYK procedure. For example, there are 6-color separations which include CMYK plus orange and green for better reproduction.
Also, if I am remembering correctly, this does not mean that yellow is only seen as a combination of green and red. There is such a thing as “pure yellow” light…ie a light from one light source with a wavelength equalling yellow. This color can be reproduced by mixing a red and green light source. (And there’s not reason it has to be red and green. You can find other shades of two colors that when combined will equal yellow. It’s just that the RGB model is able to reproduce pretty much all of the chromatic spectrum.)
Dammit, sailor. I knew I read that somewhere, but I couldn’t quite find on the Internet something stating to that effect. That’s why I wrote “pretty much all” instead of “all.” I suppose I should’ve written something even weaker like “most” or even “a lot.”
To clarify, the “dammit” was directed at myself. And seeing that thread, that’s exactly what I was looking at on the internet, the CIE chromaticity diagram. But the cites I’ve found seemed to state that most of the perceptible color spectrum can be reproduced using RGB.
Oh, I am a moron. I’ve just noticed in my history that I visited this site which clearly shows what cannot be repoduced by RGB.
Does that mean that there are many combinations of three colours that can be used either in the additive or subtractive processes in order to produce a range of perceived colours? So Printers don’t necessarily have to use cyan, magenta, and yellow as the three basic colours?
no, not really. it’s just that cyan, magenta and yellow produce the widest range of colors (hence them being “primary colors.”) choose three other colors (say, red, yellow and orange) and you’ll be able to reproduce all the hues that are a combination of these colors, but of course you’d be unable to reproduce anything involving blue or green. that said, your oranges in such a case, will be much more vibrant and accurate than anything you’d get from a CMYK process.
Certainly red, yellow, orange won’t work, but wouldn’t something like orange, purple, green also not work very well? I believe (he writes without a cite) that the three colors chosen, whether for additive or subtractive use, have to be alligned with the peaks in the color receptors in our eyes.
I’m not a color expert, more of a color dilletant. My understanding is that the primaries in either a subtractive or additive system do not need to be aligned to your cones’ spectral response curves. What they do need to have is linear independance of each primaries’ spectral curve convoluted with your cones’ spectral response curves.
I don’t believe this is sufficient. For example, Orange, Red, and Purple are linearly independant, but to get green, you’d have to have Orange + Purple - Red, and the minus sign is problematic.
If the receptors are for Red, Green and Blue, and those are the colors you’re using, it’s easy to see how to simulate any color. Instead, suppose you’re using Yellow, Cyan, and Magenta light, where yellow light excites the red and green receptors, cyan excites green and blue, and magenta excites blue and red, and try to make green. You can make light green by adding yellow and cyan, giving you two parts green, and one part each red and blue, but how do you make dark green? You’d need to subtract magenta, but you can’t subtract. This is a simplified argument, since the receptors’ responses aren’t so simple, but I think the jist of it is valid.
I was totally amazed when my question appeared in the Dallas Observer so many years ago. I was equally amazed when it appeared in Return of the Straight Dope. And now, what is really spooky is how similar your question is to the one I asked…
Zenbeam, any three colors that you pick will not be able to represent the full spectrum of light. Some choices have a larger gamut than others. As sailor said, if you look at the CIE chart (say, the one that that pulykamell linked to,) and note where your three colors fall, you will be able to attain any color that is within the triangle they form.
Part of the problem is that we talk about the red, green, and blue cones. But the “red” receptors in the eye are not red. They are greenish yellow. Here is a page that shows the cones’ spectral responses.
Sure. I don’t believe this contradicts what I said. Red, Green and Blue correspond to your color receptors, and give you the largest triangle. Look at the bottom chart on this page, and try using Yellow, Bluegreen and Red purple as your colors. The resulting triangle intersects the center region labeled “white” in the center of each leg. You can’t make Red, and Green and Blue will be very pale. The reason the Red Green Blue triangle is the largest is because that is the one which is best alligned with our color receptors.
If alignment with your color receptors was the most important thing, the three primaries would be indigo, green, and yellow. Look at this link, or the link in my above post. Your “red” color receptor peaks at 580 nm, which is yellow, not red, which itself is situated around 650-700 nm. Your “green” color receptor peaks at 540 nm, which is admitedly green. Your “blue” color receptor peaks at 450 nm, which is arguably indigo, not blue, though it’s pretty close. It is not the color receptors’ peaks which define the best primaries to pick. I’ve noticed psychophysics papers calling them the S, M, and L cones more often these days, possibly to avoid the confusion of calling them blue, green, and red cones when those aren’t their actual colors.
What is relevent, I believe, is the way the nerve cells on your retina couples neighboring cones’ signals, which they then pass on to your brain. The three signals sent are luminance, blue-versus-yellow, and red-versus-green.
Note, I misidentified the shade of the L cone in my prior post, I had said it was greenish yellow. But its peak at 580 (or 590, some sources say) nm is yellow with an orange tinge.