Color Wheels are wrong? How color vision actually works

Color theory is a little obsession of mine. You’re here for startup advice, but this week I’m taking an indulgence. Leave a comment if you want to see more or fewer of these little distractions.

Why are artists special?

Ask any artist to explain how color works, and they’ll launch into a treatise about how the Three Primary Colors: red, blue, and yellow form a color “wheel:”

PainterPrimary.png

Why “wheel?” All other colors are created by mixing these three colors in certain proportions, they’ll explain. In particular, mixing equal quantities of each pair of Primary Colors produces the Secondary Colors (orange, green, and purple):

PainterSecondary.png

Continuing this process produces the infamous color wheel you probably learned in school; a pretty, symmetrical, satisfying device in which each hue melds seamlessly and linearly into the next:

art-factory-color-wheel.jpeg

Unfortunately, none of this stands up to even minor scrutiny.

For example, open up your desktop printer and you’ll see something quite different:

toners.jpeg

Three colors of ink which, when combined, produce all others: cyan, magenta, and yellow. (Black is included as a money-saver — black is the cheapest and most common color; it’s cheaper to have a black cartridge than to dump ink from the other three.)

But wait! I thought the “Primary” colors were red, blue, and yellow, not cyan (bluish-green), magenta (bluish-red), and yellow. So this is a different set of three colors which are “Primary,” yet still generate color wheels containing all the other colors. So what does the “Primary” designation really mean?

cmyk-color-wheel.jpeg

Also it’s not as simple as saying “any three colors can produce all the others” because that’s clearly not true (by experiment). And it’s not as simple as saying “any three colors will do, they just have to be equally spaced around the color wheel,” because yellow is common to both the painter’s and printer’s wheel, yet the other two primaries differ completely (red and blue are primary in the painter’s wheel but secondary in the printer’s wheel.)

TVs and computers are different yet again. If you stand close to a CRT (non-flat-screen), you can see that every pixel (or “dot”) is really three tightly-packed colored phosphors: red, green, and blue.

crt-pixels.jpeg

If you’ve done computer graphics you’ve been forced to name colors using these “RGB color values;” true geeks automatically think “yellow” when they see #FFFF00. (If it’s intuitive to you that #A33F17 is burnt orange, you are indeed a God among men. I’m looking at you, @soopa.)

This leads to yet another system of three “Primary” colors generating all the others, and another color wheel. This one is a little easier to explain — ink and paint are subtractive (adding cyan, magenta, and yellow yields black) whereas colored light is “additive” (meaning if you blast red, green, and blue you get white):

additive-colors.png

Still, we have yet another color wheel in which two (but not all three!) “primaries” match those of the artist’s wheel and none match those of the printer’s wheel.

RGB-wheel.jpeg

This isn’t adding up. Let’s turn to science.

Physics makes it worse

Physics is clear and certain. Light is a wave of energy (or a particle, but for today it’s just a wave OK?) and, like a vibrating guitar string, light waves wiggle at certain frequencies. Some of those frequencies we detect with our eyes, and the frequency determines its color:

spectrum.jpeg

Now we’re getting somewhere! Or are we?

First off, we’ve suddenly lost the notion of a “wheel.” As much as the previous color systems have contradicted each other, at least they all agreed that hues transform smoothly and continuously, one to the next, a beautiful symmetry with neither beginning nor end.

But here we have a clear beginning (red) and end (violet). The colors in-between are continuous — and seem to generally match the order seen in the various color wheels — but then it just terminates with violet. How does it get back to red? What about that fuchsia / magenta / purplish-reddish color which is clearly present in every color wheel but missing from the physical spectrum?

How can a color be missing? Where does it come from?

But wait, we’re not done.

Another thing to resolve: Opposites

Every seven-year-old kid in America is taught that “the opposite of red is green” and “the opposite of blue is yellow.” But what does that mean exactly?

After all, there’s nothing in that linear physical light spectrum to indicate that any color is “the opposite” of any other, particularly not those two pairs. And the color wheels aren’t much help either; trying to match the “opposites” on the painter’s wheel yields an unsatisfying asymmetry where two of the primaries are opposite, and the third is opposite from a secondary:

PainterOpposites.png

But “opposites” are real. In the early 1800s Goethe (yes, the Goethe) noticed that red/green and blue/yellow were never perceived together, in the sense that no color could be described as a combination of those pairs. No color could be described as “reddish green;” if you are asked to imagine “a green with a bit of red,” nothing comes to mind. In the following 150 years, various experiments were devised to test this idea, all of which validated his observation.

There’s something to this. Something neither the wheels nor the spectrum can explain.

It’s time to get down to the real source of color: The ridiculous complexity of human beings.

The answers: Physiology (of course).

Caveat Emptor: The following is a gross and irresponsible over-simplification of what actually happens. But it’s correct in its general thrust, and few people on Earth (myself excluded) are qualified to explain with complete accuracy, so in the interest of general illumination, no pun intended, OK maybe intended just a little bit, I’m doing it anyway. So there.

Of course it starts in the eye, where three types of cells called “cones” measure the amount of red, green, and blue light hitting the retina.

“Ah ha,” I can hear you CSS freaks scream, “it’s RGB after all! I was right! All that time spent — nay invested — in knowing things like #001067 is the default title-bar color in Windows 95 was well worth it!”

Hold on there, cowboy. Actually “amount of red, green, and blue” is a gross simplification, as I warned. Peaking under the hood (just a tad), the three types of cones are in fact denoted S, M, and L for “short, medium, and long” wavelengths, and actually respond to a range of wavelengths, with a certain level of response for different wavelengths, like so:

Cone-fundamentals-with-srgb-spectrum.png

But I digress, and besides I did promise to be all gross and irresponsible, so I’ll stick with that.

So there are R, G, and B cones. The signals from these cones don’t go straight to the brain; they first pass through a pre-processing filter, and it’s this filter that explains all the mysteries. Actually there are three filters.

Filter #1 works like this:

RG-Opponent.png

Explanation: The more R there is, the more positive the signal; the more G, the more negative the signal. If there’s relatively equal amounts of R and G — whether neither of both, a little of both, or a lot of both — the signal is zero.

This explains why there’s no “greenish-red.” Because:

Let’s say R and G can go between 0 and 100 units of intensity. Consider the case of “full red with a little green,” where R=100 (full intensity) and G=25 (one-quarter intensity). Then separately consider the case of “strong red with no green,” where R=75 and G=0.

In both cases, Filter #1 computes the same output signal: 75. But remember the brain doesn’t get the raw R and G signals — it only gets the filter’s output — so the brain cannot tell the difference between these two scenarios.

So there’s no such thing as “red with a little green” — there’s just a less intense red. The brain physically cannot see “greenish-red” because the filter removes that information.

Knowing that blue/yellow is the other opposite pair, you can probably guess what Filter #2 is:

B-RG-Opponent.png

Here blue (B) is opposed with a combination of both the R and G channels. The R and G cones are stimulated either when there’s literally both red and green light (like when a CSS coder turns on both red and green as #FFFF00 to create yellow), or when 570nm light (yellow, on the visible spectrum) stimulates both R and G cones.

Filter #3 is simple:

RGB-Opponent.png

In short, it measures the quantity of light without regard to what hue it is. This is “how bright,” or “luminance” in color-theory parlance.

And magenta? It comes from full R and B with no G, activating Filter #1 full-positive, Filter #2 at zero. It’s not a physical wavelength of color, it’s just a combination of outputs from two filters.

The “Right” wheel, simplistically

To do this “wheel” thing properly, you have to represent the red/green and blue/yellow opposites. It’s not at all difficult, so it amazes me how rarely it’s seen or taught:

4-color-wheel.gif

Four primary colors? Yes, why not? It’s the closest thing to the actual physiology without getting complex.

Bonus Brain Bender: The context / color connection

This is just the beginning of color theory. To give you a glimpse of how complex it gets, consider this:

When a color is juxtaposed with other colors, we perceive it as a different color. For example, most people will say the small square on the left is orange, whereas the one on the right is brown:

albers_interactionofcolour.jpeg

Actually, the squares are exactly the same color! The surrounding context dictates the perceived color, on top of all that wavelength-physiology we just did.

It gets worse, because the brain projects abstract things it knows about the natural world onto your perception of color. For example, we know intuitively that shadows artificially darken colors, so our brains automatically account for this in our perception of those colors. (It’s called “color constancy.”) For example, you know that the dark and light colors on this hot air balloon are “the same:”

Hot_air_balloon_-_color_constancy.jpeg

But it also results in optical illusions so powerful that even when you know the trick you still can’t see it correctly. Like this: Which square is darker: A or B?

Grey_square_optical_illusion.png

In fact A and B are the same color (#787878), but you can’t see it even when you know this. To prove it to myself I had to open this picture in an image editor and actually move one square over another to see it was the same.

Freaky.

Further Reading

You got this far? You still care? Sheesh, you’re as weird as me.

If you really want to lose a few days of your life, this is an amazing, in-depth treatise on color theory. That link is just page 1 of 8. Good luck.

165 responses to “Color Wheels are wrong? How color vision actually works”

  1. the trick of the blank square / dark square is astonishing… I also checked in my imaging software. the orange square would not trick my (computer scientist :) eyes… but this one, wow.

  2. Not the type of thing I usually open google reader for, but, well… I guess thats what made me read it till the end (apart from being interesting). Keep it coming.

  3. Perfect timing — my 5th grade daughter had vague ideas about doing something with color perception for her science fair project and I told her that I’d need to learn more about the topic to help her out — this is enough to save at least one trip to the library.

    (Now, if you can just craft an experiment for her, too….)

    • 1. Make some 4″ circles out of posterboard.
      2. Paint each side with 2 colors that when blended produce a 3rd color –
      like red/yellow = Orange. blue/red=Purple for instance.
      3. Poke holes on the sides of the circles.
      4. tie loops of string to the holes.
      5. put a hand in each loop and spin the circle until the strings go taught.
      6. pull outward on the strings.

      As the circle spins ( flips over and over ) quickly, viewers will “see” the color that the 2 colors on each side produce together. When the spinning stops, the color is gone.

      Good luck.

  4. This is the best blog post I’ve read on colors, and what they mean to humans. I tracked photons on their trail across galaxies, but your post digs into the human filter aspect.

    If we dig a little further into human color perception, we can see (pun intended) that those cones pick up light with rhodopsin, and that gene leads to signals in our optical nerves which can be abstracted to those filters.

    Hey Jason, try out disqus. It saves me from having to type in my name, email and website :D

  5. Easy to find experiments — just look up optical illusions! It’s fun too.

    A common one with color “opposites” is where you stare at an image for 30 seconds then look at a white background and the colors are inverted.

  6. If you really want to do something fun, try rendering an image using SML colors. In Ray Tracing, this produces very realistic images.

  7. Hilarious!! Thanks!

    Especially as I was just trying to work out what colours to use for my new logo… now I know… get someone else to do it!

    ……… and tell them I need a purple with a little bit of blue in it!

  8. At the risk of sounding completely silly next to all you techy types….
    .. but the only silly question is the one you don’t ask.. right?

    How do you put your picture up??

  9. Have you ever heard of the term “polymath”? It’s when someone excels at a broad range of completely unrelated topics… and I think this is a fantastic example of how entrepreneurs can be passionate about not only business but other things such as science and art. Should you post more articles unrelated to business? I say definitely. An interesting person is a person interested in the world around him/her. I think it can only make you better to know more and take interest in more things… as an entrepreneur just like as a business person in general or as any person in any field!

  10. Coloring outside the lines is expected: More of this please. And take heart, you’re in good company. Check out the lifework and hobbies of chemist Wilhelm Ostwald. Thanks for the education.

  11. As another technologist and armchair color theory junkie, I appreciate digressions like this. More please, Jason!

    …Just don’t get me started on the subject of fonts and type, or I’ll lose hours of productivity….

  12. You got the contrasting (opposite) colours confused. The are red – green (right) and yellow – purple and blue – orange (not blue – yellow) yeah artist head here :)

  13. Hi Jason,

    Really interesting post. Now I’m going to have to look up how ink jet printers produce their colors versus artists paints – suspect it has something to do with dyes versus pigments.

    Another color area you may want to investigate further involves gemstones. A fun example is that chromium produces both the RED in rubies (corundum AL2O3) AND the GREEN in emeralds (beryl – Be3Al2SiO6). Tourmaline is another interesting stone because it can be one color on the C axis like green and another color on the AB axis like red. Fun stuff.

    Thanks.

  14. You apparently have never seen a Vectorscope. In your “Right Wheel” example, you show what is essentially a vectorscope turned on it’s side.
    Vectorscope’s are used in monitoring file based, tape based, and film based broadcast media for transmission or recording. What you’ve “discovered” here is that there are in fact different color “spaces” that are dependent on the intended viewing platform. Print ends up using CMYK colors, Video uses RGB. Mixing paint additively usually refers to a standard Artist color wheel.
    Color correction devices use a 360 color sphere that mimics a vectorscope. If you went digging for the requirements of video or film projection, time square billboards, high gloss coated banners, billboards, stadium monitors, web banners, etc…. you would find different intended color spaces, and different display characteristics requiring adjustment to the Color Space used to create the intended viewing material.

    This gives a very basic idea of color wheels and vectorscopes for Video color correction:

    http://www.kenstone.net/fcp_homepage/fcp_7_scopes_vectorscope_stone.html

  15. More distractions please – this is fabulous stuff!

    Also, on a completely unrelated note, is it possible to use Disqus for comments? It’s unbearably more convenient, just in terms of being able to track all your comments in one place …

  16. Awesome article!

    I’ve heard of people being able to hack their optical hardware to be able to see reddish-green and bluish-yellow, though I’ve never gotten it to work for myself. Allegedly, there are another set of ‘filters’ that work from the data provided by the ones you mentioned and give rise to our actual perceptions, and by doing weird things you can see ‘unseeable’ colors.

    http://io9.com/5710434/train-yourself-to-see-impossible-colors

    The original article I saw was in Scientific American – if you’re a subscriber, it’s here: http://www.scientificamerican.com/article.cfm?id=seeing-forbidden-colors

  17. Phew been in colour all my working life. I am a CYMK sort of guy and did enjoy the take on the colour wheel, Nice work.

  18. Do NOT stop putting stuff like this out…I have always wondered about this, and never thought to put in the time and effort you did to explain it so exceptionally well…Kudos! Love it! thanks Jason…well done!

  19. so you’re leaving out a major point… there is a huge difference between adding colors and adding ink. When you mix ink, you actually remove color. For example, if you mix yellow and blue ink, you are left with green. What’s happening here is the the blue is absorbing some of the reflected light from the yellow, and the yellow is absorbing (removing) some of the green.

    When adding color, as in light (what you learned from the color wheel, and your rgb monitor) you get more light with each color. The point is, if you shine all wavelengths of color you have white light, if you mix all inks, you have dark black (everything is absorbed, nothing reflected back).

    • If you’ll read again you’ll see I did not, in fact, leave out this point. And that point doesn’t solve any of the problems with traditional wheels.

  20. Very interesting. A related question: If you wanted to pick N colours that were as distinct from each other as possible, how would you do it? Using a colour wheel? Bonus credit for making them as distinct as possible for people with the most common types of colour blindness!

    • So first you wouldn’t want to use any of the color wheels here. Rather you should use a color space specifically designed to model human vision, e.g. Lab space.

      There’s also “different” in hue versus other components like how much white or black is mixed, and overall luminance.

      For differences that account for typical forms of color-blindness (e.g. green/red), there actually exists an entire theory and recommendations and color spaces just for that.

    • I did a project on this topic. What you need to do is take a perceptually uniform colour space and draw the largest possible circle in it that you can fit, in three dimensions. Then, take colours along equally spaced coordinates around the circle. The catch is that the circle has to be far away from the background colour that you are using, too. The LAB or LUV spaces work well for this.

      • >draw the largest possible circle in it that you can fit, in three dimensions.

        Circle or sphere?

        >Then, take colours along equally spaced coordinates around the circle

        Around the circumference, but not inside? Why not inside?

      • If you mean “distinct” as in “RGB values most different,” then yes. It’s easy to do that — just use the HSV model and create equally-spaced hues around the circle.

        If you mean “distinct” as in “human perceives the colors as appearing as different as possible,” then no. You need to do what Steve Hanov proposes and use a colorspace like Lab.

  21. Fascinating stuff and a good read Jason! Now I just need to figure out how to use this new data to make better color decisions for apps, web sites, photography, and so on. Time for some experiments…

  22. This post clears up a few misconceptions for me, thanks for that.

    I think you will enjoy the following link about “context” related illusions: http://www.psy.ritsumei.ac.jp/~akitaoka/color12e.html

    Akiyoshi Kitaoka is a university professor from Japan that has created a very extensive and amazing collection of illusions, documenting so many subtle ways our eyes play tricks on us.

    His homepage is really worth exploring:

    http://www.psy.ritsumei.ac.jp/~akitaoka/index-e.html

  23. Great article – I’m sure I’ll refer to this as I write the color part of my book, “Design for Hackers.”

    The one part I’m confused about is saying that blue and yellow are opposites. In years of art and design training, I don’t recall ever hearing that. Instead, orange is considered the complement to blue, while violet is considered the complement to yellow.

  24. Nice post.

    I’m not sure if it’s right, but I was taught that there is a difference between the additive and subtractive color systems. So, where we have light sources combining, like a monitor, it is additive and breaks down to RGB. In print however, the ink subtracts, so it is CMYK, where the K is black. Another small point that may or may not be correct is that black is added separately because it is difficult to get a good solid black from the combination of CMY. In some printers I’ve even seen multiple black cartridges that get used for varying effects (epson 3800 for instance). Also, there are other color systems and a multitude of issues that involve gamut (but lets not go there :-)

    Paul.

  25. Thanks for the article. I studied a lot about the color theory when I created my color application, and I think I understand why the classical arts used the RYB wheel even if—from some points of view—it might seem yellow–blue should be the opossite colors. But they aren’t perceived as opposites, in fact.

    You can try a simple experiment: if you stare at a solid color area for a while, and then you close your eyes, you will “see” the opossite color. If you stare at yellow and blue, you won’t see blue and yellow. Your own eyes can tell you the opposite for yellow is purple, and the opposite for blue is orange. You can try it directly on this page: find this picture above, the one with yellow and blue stripes in the middle and those orange squares on both sides. Lean to the display so that your eye distance is about 20–30 cm (up to one ft) and stare at the picture for 30–60 seconds. Try to keep your eyes as still as possible. Then close your eyes and relax. You should see the image on your retina in inverted colors. Instead of blue you’ll see orange, and purple/violet in place of yellow.

    There is also a practical reason to deny the 4-primary-color RGYB model. Often, as an artist, you use not only the opposite color to make a contrast, but also surrounding colors to make lighter, softer contrast. It works great with the RYB wheel: for blue, you can pick either the opposite orange to make the strongest contrast, or surrounding colors from red to yellow for softer accents. The effect is pleasant and it’s perceived as nice color combination – i.e. like this: http://colorschemedesigner.com/#3L31Tw0w0w0w0

    In the RGYB model, it works the same for red and green. But if you pick blue and use yellow as the “opposite”, you shouldn’t pick colors from both sides of the yellow. You will get unbalanced and often unpleasant combinations, like blue+orange+green, which are more agressive, not recommended by classical color theory to be used together (unless you are creating something agressive on purpose). You should choose only one side: either use the yellow-orange side, or the yellow-green side. You should avoid using a blue color with both yellow-orange and yellow-green accents together.

    For the same reason both the RGB, and the CMY wheels fail with geometry used for pleasant color combinations. The classical RYB color wheel is thousands years old and used by artists through millenia. There are many good reasons why the colors are organized exactly this way. Even if physics sometimes says otherwise.

    However, many thanks again for the physical and physiological background, it’s very useful and interesting reading.

    • Thanks for your comment Petr! For further insight into the perspective of artists and those practiced in color theory, Josef Albers’ “Interaction of Color” and Johannes Itten’s “The Art of Color” are required reading.

  26. Cool info about the human perception bit — didn’t know that!

    Related to printing colorspaces, you should look into hexachrome printing. More “primary” colors give the printer a wider range of reproducible colors.

  27. Just when you think you may understand something the reality of the human body/brain throws it for a loop. Still… nice to have the perception of color.

    That last A/B square example is amazing.

  28. This is by far the greatest article I’ve ever read about this subject. And I read many, but could never really wrap my brain around the different aspects. Thanks!

  29. errrmmm… I don’t teach color theory the way you suggest artists explain it. I teach pretty much everything you’ve covered in this post, plus some of the physics involved with light absorption and reflection, physiological and neurological functions of the brain/nervous system/ eye. And the cultural and historical differences in how color is perceived.

    The color wheel is simply a shorthand way to discuss the subject without getting all long winded and egg-heady with people who aren’t really interested in the intricacies of color theory.

    Don’t be hatin’ on artists.

    • You’re right, it’s unfair to lump any group together. But I have, in fact, had this conversation with dozens of artists, some of whom do it as their primary living, and I’ve never once heard anyone be able to articulate this stuff, and most are stuck on the painter’s wheel with no comprehension of why.

  30. Great stuff, I really like the idea of 4 primary colors. I still can’t understand were the grey scale is formed (nor the browns)

    • So this post was mostly about hue rather than white/black mixtures and luminance.

      The greys (including white and black) arise from a (relatively) even application of wavelength in the visible spectrum.

      Browns are tricky too. Brown is dark orange and context. As in the example above where both squares are “orange” but one looks brown.

  31. You say “Every seven-year-old kid in America is taught that ‘the opposite of red is green’ and ‘the opposite of blue is yellow,’” but this is wrong. In the painter’s color wheel the opposite of blue is orange and the opposite of yellow is purple. Didn’t you notice that the line you drew between blue and yellow had to bend in order to connect them?

    • Those are “complementary,” meaning they oppose on the painter’s wheel, but they’re not physiologically opposite, which is my point.

      • Actually, yes they are. We can prove this quantifiably because if you mix them together you get a shade of gray. If you mix any other two colors together that are not complimentary, you will get a non neutral color. There are no exceptions to this.

        Repeatedly, in the article you mention that red and green are opposites because there is no such thing as red with a little green. That is absolutely correct. However, here is where the problem comes in. You also repeatedly claim that blue and yellow are the other opposite pair. Okay, using that logic, there must be no such thing as blue with yellow. Oh wait… there is. It’s called GREEN.

        • No, you’re confusing mixing paint with mixing light, and you’re
          confusing perceptional differences with manipulating physical things.

          On Monday, February 28, 2011, Disqus

  32. Hey there. I liked this little lesson on color. Have you read about different animals that can see more colors than we can? Freaky right? Apparently there are some birds and fish like that. Anyway, just thought it might add to your wealth of knowledge. Thanks!

  33. Long ago I was digging into this stuff and had some neat thought experiments.
    If a Red/Green colorblind person identifies colors that a “normal” person identifies as different colors, could an alien (someone with different eye physiology than humans) look at what we say are “the same color” but in fact appear as different as Red/Green do to us? (and if so, what would that physiology look like… four types of cones? greater separation in the peaks? different “filters”?)

    • Yup! It’s sort of like trying to think about what the 4th-dimension looks like — you can understand the question but you can’t really “see” it.

      BTW some women do have four cones, the fourth overlapping in the orange spectrum.

      • Fantastic article. I’d never heard about some women having 4 cones! I wonder if this explains why women seem to like pink:P Maybe they experience it differently from men.

  34. Some guy claims he can see reddish-green: http://groups.google.com/group/sci.lang.japan/browse_thread/thread/65cd8c2c0b521fe1/ac0bb339db0448e6?lnk=gst&q=eicher+reddish+green+paint#ac0bb339db0448e6. What you’ve posted here seems to leave no possibility for that, but he also claims MIT and Harvard back him up. He cites Retinex theory, but I can’t find anything on the interwebs that links Retinex theory to the possibility of perceiving a red-green colour.

    • That guy seems like kind of a crackpot. He’s much more concerned about insulting his interlocutor (making offhand comments about architects’ lack of authority to discuss color, telling him he can’t read, suggesting he might just be colorblind despite clear evidence to the contrary in the other man’s emails) and insisting on his own correctness than he is about making a comprehensible argument. He throws out supposed supporters (Arnheim! Edwin Land!) without explaining how their models have anything to do with his point. He doesn’t bother to try answering the main questions, which are: (a) can this “reddish green” color be described any better than just “reddish green” – how does it differ from yellowish brown for example – and (b) if your visual system contradicts what we understand about neurophysiology based on both physical and psychological experiments, what explains your uniqueness and why aren’t you running to a nearby lab so they can publish such an amazing discovery?

      He might be a decent painter, but I really can’t take anything he says about the science too seriously.

      • Jacob’s right, and also again in this article I intentionally glazed over details of the science. Also cone-sensitivity curves vary between individuals — some people (mostly women) even have four types of cone! — so it’s possible that some people can see other colors.

  35. Hi,
    Nice article.
    I remember i learned that the opposite of blue was orange and the opposite of yellow was purple, and red versus green, so that all opposites are secondary. That is making sense about the mixture part because you can’t have purplish yellow but you can have green as a combination of yellow and blue. As to remember these “Ghoethian” theory was to stare long enough to a simple form in a primary color, e.g. a yellow dot on a white background, and than look beside it to the white surface, where you could see the same form in its opposites color, purple.

  36. Just a side note: Haircolorists are taught a slightly different color theory than the ones you mentioned as well. In this version, blue and orange are “opposites” as are red/green & yellow/violet. When combining “opposites” you get a form of brown, hair’s “neutral” color.

    Haircolor pigments are not meant to be mixed as “equal parts” to get any particular result, they are weighted based on how dark or light the color should be in the end result, as well. The more dense the mixture, the darker the end result will be.

    I’ve definitely oversimplified the process in my description – but that is the gist of it. :)

    • That’s actually exactly how the real color wheel works for artists (and everyone). They are extremely mistaken above when they say blue is opposite to yellow. It is exactly how you described it.

  37. Do you actually paint? How many of these “artists” do you know to arrive to such a conclusion? Painters use the RYB system because it works with the kind of pigment that is used to PAINT. That first part is totally unnecessary, you can explain all that comes next without that intro bashing something you don’t even do.

  38. I am a college student double majoring in art and psychology, therefore I think I am more than just “well-versed” in the concepts of color perception and color theory. Let me therefore say this: this is all a load of crap.
    You are completely wrong on almost all fronts.
    Nice try, but no. The original color wheel is correct, and has remained the same for a reason.

      • This “new” color wheel and the idea that blue and yellow are opposites and they cannot be mixed to make any discernible color is completely false, on both counts. In fact, that blows this new color “theory” completely out of the water. Blue is not opposite to yellow and if you mix them together, you get green. That is 100% fact.

  39. I’m glad for any color wheel that eliminates the “yellow complements violet” component, because to my eyes, those colors would be the worst combination ever in design. I don’t think I’ve seen yellow and violet work well together, ever. :)

  40. Great illustrations, and well put!
    I grew up with the NCS color system, which uses the four primary colors at the end, so I’ve always been annoyed at the “three color” approaches.
    (I think they are on the web these days: http://www.ncscolour.com/ )

    • I´m amazed that not more people have mentioned the NCS (Natural Color System) – I thought it was a big part of the curriculum for anyone interested in color?

  41. Very interesting!

    Before reading the end of your post I was already checking in Photoshop whether those two squares are the same colour.

    I still cannot believe it!

  42. The main point of our color perception system is to maintain a consistent color “name” no matter what color light is hitting an object. If you see a tiger under a forest canopy (green light) and don’t recognize it as a tiger, you get eaten. The god of color vision theory is Edwin Land (of Polaroid fame). His Retinex theory of color perception was the first to successfully explain a number of visual illusions ( the full color hamburger photo make up of only red and white light for example). We have a full color electro-optic system that uses only two channels to create all the colors and neutral grays for the user. Possible only with Retinex as the explanation

  43. Thanks for putting this together. I write color correction software for film restoration, converting pixels between color spaces, applying filters…
    The brain model you present here helps a lot to (begin to) understand how we perceive images.
    Thanks again!

  44. I ain’t one of you smart heads, but I do some painting and know our wheel works for us because opposite colors produce gray. (r+g b+o and y+p). It is a great tool for rendering object to use the opposite to darken instead of just adding black. With your wheel y and b would make green which is the other primary color. The real problem is that they imply that colors are equal, but some are much stronger that others. I would like to see a wheel that has color that mix equally. example: if you mix equal parts of red and blue it makes a very dark purple that is nearly black. Where is brown in all this? Very interesting article though. I will try your wheel at my next session.

  45. Jason, I was mainly impressed by how clearly you developed your explanation. That alone kept me reading to see how you would handle each section. Well done.

    On the checkerboard, where the shaded white square had the same brightness as the exposed black square: a sheet of black paper in bright sunlight reflects more light than white paper in a very shaded room… but the white paper still looks white. As with our pre-perceptual color processing apparatus, we also have neural apparatus for automatically comparing the brightness of adjacent objects, relative to each other. When the black paper and the white paper (inside) are tested by light meter, there is no relative judgement. The numbers indicate actual intensity rather than relative intensity.

  46. Great article, thanks. As an exercise i am trying to work out the settings for some ‘Primary’ hues. if a magenta hue arises from a Filter One value of full positive, Filter Two at zero – what kind of hue would be FilterOne=0, FilterTwo=fullPositive ? Also if you could clarify the filter values for a pure red, green, blue and yellow i would really appreciate it, thanks.

  47. Can’t believe the art freaks still don’t get it even after reading your post. I’ve always believed (even since little) that pigment wheels were incorrect. Light is the true way, but always wondered why yellow wasn’t included. That was a long time ago and thanks for writing this and giving me good times.

  48. I don’t understand your reasoning for 4 primary colors. You do not explain how white and black are possible or why we can see them on a monitor. I started a web color test which I ended up showing on a hexagonal triangle grid since I realized that color could be expressed in a cube. I was to discover later that this was sRGB colorspace. Here is one example using CSS animation (view in Safari 5).

    http://css-class.com/test/css/3/transform-color-cube3.htm

    Here is another version that reveals the planes through the cube. Each plane has two hues (one positive and one negative).

    http://css-class.com/test/css/3/3d-animation/color-cube4.htm

    The more saturated colors are further from the white to black axis. Lightness is different from these X, Y Z axises and is more the way the cube is orientated.

  49. It’s nice to see the scientific principles behind color TV engineering explained in a commonsense way. Now we know why “component” connector cables come in sets of three. See this from Wikipedia:
    “YPBPR is converted from the RGB video signal, which is split into three components, Y, PB, and PR.

    Y carries luma (brightness or luminance) and synchonization (sync) information.
    PB carries the difference between blue and luma (B − Y).
    PR carries the difference between red and luma (R − Y).

    Sending a green signal would be redundant, as it can be derived using the blue, red and luma information.”
    So we’ve known this stuff for a while – too bad it can;’t be taught in school

  50. The most significant work with colour was done by Isaac Newton. By demonstrating that a prism could decompose white light into a spectrum of colours he unlocked the only real science of colour. Since then man has tried to determine, categorize, and limit colour to a single method, system, or chart.
    Artists, architects, designers have argued, discussed and dissected the various merits of the colour wheel – yet colour is lineal.
    The language of complementary, split complementary and the vast majority of colour theory are built around the notion of a colour wheel – a falsehood, a convince that has no place in nature.
    The charts of the last century are just as random– they merely provided a common language to describe a subjective experience. They represent the ideas and culture of the men and time that produced them.

    • That is factually incorrect. Newton himself invented the first color wheel which he promulgated in his magnum opus Optiks in 1704. More info here.

      • “After Newton had used a prism to separate daylight and count seven individual colours, it appeared to him that, when considering colour-hue, this was a closed system. By taking the violet end of the spectrum and linking it to the red start-point, he thus created a convincing circle of colours. With Newton’s circular shape, the transition between the one- and two-dimensional colour-system is complete. It is helpful to realise that although this step was made by a physicist, it actually has little to do with physics; it is our brain that, out of the straight line of physics, makes the circle first drawn by Newton” -from http://www.colorsystem.com which is a wonderful to look at mans attempts to develop a universal Colour system.
        Sorry, not sure which bit is factually incorrect. It is not in dispute that newton developed the first wheel based on the results from his work with the prism rather the stars and planets(though the idea of a circular chart had been around for a long time). The point I was trying to make (badly obviously)and one that your link supports is that colour from a light and physics point of view ie Colour as it is actually ordered is lineal – newton ‘bent’ it into a circle because he viewed it as a closed system. A circle certainly was an obvious choice but perhaps not the only one. It is because newton choose a circle that we use the colour wheel the way we do. I have found no evidence that he believed/supported the idea of complementary or split complementary color- or that this had any bearing on his work.

        “Newton’s colour circle will remain inadequately explained if we ignore its inventor’s belief that the propagation of both light and sound are comparable, and that they should therefore be treated harmonically in an identical way. Newton selected his seven colours because an octave displays seven sound intervals. He allocated segments to them in accordance to their value in the Dorian musical scale. The individual sound tones associated with this scale coincide with the borders between the colour grades: D, for example, with the border between violet and red; A with the border between green and blue. This mathematical-musical appropriation of colours makes it difficult for many to understand Newton’s system which, with its seven (instead of five) primary colours, has more of an aesthetic basis than a scientific one.” http://www.colorsystem.com That last line is my rather muddled point.
        Regards Rachel.

  51. When I was in university I noticed that there is a language gap: when an artist says “blue, red and yellow”, he is actually meaning “cyan, magenta and yellow”.

    (sorry for my poor english ; )

  52. This article was a great deal of work and has generated a lot of comments. It even had me fooled into second guessing what I’ve known to be true about color from working with it for over 20 years. Unfortunately, it makes a huge mistake, as others have pointed out – the complement of yellow is purple, not blue, and so the “traditional” wheel that is dissed here is not incorrect. Complementary colors “complement” one another by essentially clashing. Complementary colors are used to create visual impact, not to be harmonious.

  53. In reading over your article, one thing I think would help you understand color wheel application is the idea of emitted and reflected light. The reason printers use CYMK as a process is that light is reflected from the paper to your eye. None of the colors in your deskjet (or from a commercial litho press) are opaque – rather, they are transparent to let light pass through to the paper below and ‘bounce’ back to your eye.

    When artists speak of the color wheel, it is a shorthand way to arrive at a color solution, but if you’re trying to be specific one would have to also address saturation, value and intensity (as well as color relationships, which you touch on later in your post). In other words, paint and inks are both material – pigment and vehicle mixed together. If you mix yellow and blue, you get ‘green’. But that ‘green’ would be darker in value and of lower intensity than any paint made from a green pigmentation alone (such as sap green, or Hooker’s green, for instance).

    I’ve taught color theory for a long time and I think it’s important to actually mix colors, take notes on the results and to apply that knowledge to the project at hand. The color wheel is not intended to be authoritarian. You might find it useful to search for information on color trees as opposed to wheels, which is a closer representation to how a painter uses and conceives of color.

  54. Nice, I love discussions that combine physics and physiology. I saw a talk recently by Dr. Mitchell Feigenbaum called Looking and Seeing. It was about how optical effects combined with human visual sensory characteristics can create situations where something looks very different just by small changes in perspective like the viewer tilting his/her head sideways. The key example was astigmatism in optical wavefronts.

  55. Nice, I love discussions that combine physics and physiology. I saw a talk recently by Dr. Mitchell Feigenbaum called Looking and Seeing. It was about how optical effects combined with human visual sensory characteristics can create situations where something looks very different just by small changes in perspective like the viewer tilting his/her head sideways. The key example was astigmatism in optical wavefronts.

Sign up to receive 1-2 articles per month: