Heya smart folks. I hope you can help me out here, this actually kept me up all night.
The colors we see are wavelengths of light in a spectrum, right? At one end of the spectrum we see red, and at the other end, purple. If you go any further past these extremes, you quickly get into invisible wavelengths.
So how does a color wheel work? How can blend purple into red without passing back through, say, green and yellow and so on?
I would imagine that, if you measured the wavelength of light reflecting from a color wheel, you’d see a gradual change down the spectrum until you hit some specific spot between red and purple, when wham , it jumps clear across the spectrum. Would it actually do this? Or is there a gradual shift from purple to red that doesn’t include the rest of the colors in the spectrum?
The color doesn’t follow the spectrum as such. It merely arranges the colors in a logical order based upon what goes with what and what is complimentary with what. Also with what we see. It only mostly follows the spectrum.
Purple is what we see when we combine blue pigments with red pigments because for some reason that is how our brains are wired. It overlaps the two opposite ends of the spectrum, rather than allow there to be any ‘holes’ in our perception of color. The spectrum does not in actuallity overlap, but our minds do it as a short-cut.
There are in fact 2 color wheels. One for reflected light, and another one for emited light.
The reflected light one is the more familiar to most people. It is used for things like paint that do not emit light of their own, we only see them by what colors they reflect. It has 3 primary colors, red, yellow and blue, and their compliments, green, violet and orange, placed opposite them. The compliment of a color is a color that when mixed will produce grey. Purple is opposite yellow because they are compliments, and being opposite yellow places it inbetween red and blue. Our brains must be wired to do this automatically as well, allowing red to fade into blue in perfect compliment to the way the colors on the other side of the wheel ( with DO aloign with light frequencies) blend into each other.
The other color wheel is the emited light wheel. The primary colors here are red, green, and blue. The complimentary colors here are cyan, magenta, and yellow. Complimentary colors on this wheel add up to white. Adding yellow light to blue light in roughly equal porportions will create what appears to be white light. You can also add all three, of either the primaries or the compliments, to get white.
So, to sum up, I think we see red fade into blue through purple is because the way our brain sees color is wired so that compliments fade into each other the same way. So red fades to blue as green fades to orange.
There is more than one way to model what we percieve as color and just looking at the visible spectrum may not make the whole thing apparent. Did you ever notice there is no brown in a rainbow? You won’t see it if you look at a one dimensional color model. If you model color with three dimensions you have a lot more to work with. You can use the additive red/blue/green model of display devices or the subtractive cyan/magenta/yellow/black model of color printing plus the the hue, saturation, luminance model.
Brown is simply what you get when orange is partially neutralized by its compliment towards grey. Any painter will tell you this. You don’t see brown in the rainbow because each color is pure.
The chart linked by Q.E.D. is an emitted light color wheel, usefull for things like television sets.
I think it is also important to note that even if there is a full spectrum in the color wheel, our eyes have only three different types of color receptors: one type detects colors surrounding red, one type detects colors surrounding green, and one type detects colors surrounding blue. So when you are looking at the yellow part of the rainbow, the light being detected is a pure yellow wavelength (around 0.55 micrometers), but your red cones and green cones are both getting excited. The color you perceive results from the relative amounts of excitation for the red and green cones.
When looking at a yellow spot on a computer monitor, on the other hand, the light you are detecting is made of a pure red color and a pure green color (i.e. RGB = (1.0, 1.0, 0.0). This is actually quite different from the pure yellow coming from a rainbow, but it excites the red and green cones in your eyes in much the same way as the pure yellow from the rainbow, so it looks like yellow.
That doesn’t quite explain why we see a continuous blend of color from red to blue when in terms of wavelengths of light the two colors are nowhere near each other. I think the way we perceive secondary colors to be complimentary to primary colors is the key to this. Since the primary colors blend smoothy, their compliments must as well to keep our internal color logic consistant. Purple light is nowhere near red light, yet our minds bring them together because yellow light IS next to green light.
On preview, I’ve decided to add a quick summary to actually answer the OP. The reason Red and Blue can blend together through Purple on the color wheel is because Purple on the color wheel has no actual relation to Violet as it is seen in the rainbow. Instead, Purple on the color wheel is defined exactly as a mix of Red and Blue. This was (perhaps) alluded to by Padeye. See below for an actual explanation.
I am now greatly intrigued by this. The key here is to determine what we mean by purple light. If we’re looking at purple light as produced by a computer monitor, it will obviously look close to both blue and red, because on a computer monitor, there is no way to produce a color like purple without actually blending red and blue light. I think this is probably quite the same with printed materials, though the base colors are different.
After a bit of searching, I find that all the colors can be represented by a chromaticity diagram. A color gamut is a set of colors from the chromaticity diagram that can be represented by the addition of different colors. Due to the nature of the chromaticity diagram, no matter what colors you are adding (i.e. the color system, e.g. RGB, CMYK [cyan, magenta, yellow, and black], etc.), there will be colors that you cannot represent. These colors fall outside the color gamut of the color system you are using.
In the case of true violet, which is past blue in the spectrum, this color does not fit into either the RGB color space used by computer monitors or the CMYK color space usually used in printing. True violet (that is, llight with a wavelength of 400 nanometers and above) can only be approximated in the RGB color system by combining red and blue.
Now, I can’t rightly say for myself if true violet as seen in a rainbow actually looks close to red (though I’ll try to remember next time I look at a rainbow). However, I can say that a good purple, as represented by RGB (0.5, 0.15, 0.65), is by very definition close to both red and blue. This is going to be the case when talking about any printed colors or any colors on a display, especially any representation of a color wheel.
A side note: with Microsoft’s wimpy Paint program (found under the Accessories menu in the Start->Programs menu), you can use the Edit Colors dialog to experiment with different mixtures of red, green, and blue.
A good website that demonstrates the chromaticity diagram and color gamuts can be found at What is CD.
The reference I used for this is Digital Image Processing Second Edition by Rafael C. Gonzalez and Richard E. Woods. This is a very good textbook for anyone interested in digital image processing (what else?), and also has some good primer information on how our eyes work and such. You can order it at amazon.com (or find it in a good academic library, maybe).
About continuity of the colour wheel… it´s not really continuous in the sense that some colours have a wider perception span, for example greens are about four times wider than violets; any reason why perception it´s like that?
I have a pet theory about that, we evolved to have a larger perception of greens, because vegetation is green, and while moving through it it´s better to have more “colour bandwidth” to gather as much information as possible, if the colour bandwidth of greens would be as thin as violets why wouldn´t be able to tell readiliy the difference between different foliages, for example. Of course all that refers to some remote time when we were still hominids.
Does that make any sense?
It’s more complicated than that. We have three types of light receptors, total, including the black-and-white rods (incidentally, this is why the space of colors we see is three-dimensional). That only leaves two color receptors. If I recall correctly, one of the types of cone is excited by red but surpressed by green, while the other type is excited by blue but surpressed by yellow. A system with red, green, and blue receptors (but no rods) could, admittedly, span approximately the same color space, but that’s not the system we actually use. In fact, any three receptors with independent response could span qualitatively the same color space, though not necessarily with the same metric.
Do you have a cite for this? Everything I’ve read indicates that normal humans have three different types of cone photoreceptors (for example, see here and here (PDF–see slides 11 and 12)).
No. We have 4 types of light receptors in our eye if you count rods plus 3 cones. Rods do not provide color information however. Their peak sensitivity is to yellow-green light.
There are a couple reasons why we perceive greens and yellows to be brighter. One is the fact that the various color receptors overlap there. So you get more receptors firing. Also, the rods are sensitive to this area of the spectrum primarily, and will make something emitting here seem brighter. The two effects combine to make yellow and green things that are relatively the same brightness as measured by a photometer appear brighter than something that is any other color.
Violet is not perceived as bright because it is only detected by the tail end of the sensitivity of the blue receptors. It simply does not stiumulate our eyes that much.
Since it is now daylight, I just took a prism and studied the purple part of the spectrum. It looks identical to the purple prduced by the RGB color space. It’s a more bluer side of the violet produced by an even mix of red and blue, but it’s still acheivable within the color space ( as tested and compared in photoshop). As far as our minds are concerned, the color produced by a 2-1 mixture of blue and red light is the same as the color of light at the purple end of the spectrum.
Seeing this, my original explaination still stands. Our mental color space is simply a circle ( well really a cylinder if you add white to black), with violet bridging the gap at the ends of the physical limits of our ability to see, and the colors arranged such that compliments blend to grey oppositely. Blue blends into violet and then into red simply because that is what the color space demands.
The color ‘red violet’ does not exist as a pure frequency of light in nature. It only exists in our minds as a consequence of our 3 color circular color space. There is no way to make a laser that will read as reddish violet to our eyes with one frequency. (Though, if we were able to see ultraviolet light, this color could be assigned to it without having to totally rewire our brains. Ditto for infrared.)
Almost, but not quite. We have two kinds of light receptors, rods and cones. Rods are sensitive to low light conditions and are essentially produce black and white vision. Cones handle bright light and come in three flavors; long, medium and short wavelength (roughly red, green and blue, respectively). These three cones are ‘wired up’ into three channels; a lightness channel (red AND green), a red OR green channel and a blue OR yellow channel (blue OR (red AND green)). If you have access to something like Photoshop, you can see how this works by converting an image to LAB (lightness, red v green, blue v yellow) and working the sliders to see how we process color physically.
