Talk:Color/Archive 1

Apparently pre-2003 contributions
White and Black are not truely colors, but the presence or absence of color.

When someone asks you what color snow is, you say white, not none. Black, white, and gray are not hues, but I think it is hard to say they are not colors.

The definition here is good, but the rest of this article (including the silly semantic dispute above) seems uninformed, culturally biased, and generally useless. The individual colors should have their own top-level pages as well, with dictionary-like entries and perhaps some coverage of cultural associations and specific uses.

A more thorough treatment should mention color spaces and standards, additive and subtractive mixing, digitization, optical effects, vision and perception in other animals, etc. The one standard reference on the non-biological science that everyone cites is The Reproduction of Colour by Dr. R. W. G. Hunt. Perhaps someone else can recommend a reference for the biological end of things? --LDC -- Is cyan (500 nm) a "familiar" spectral color in other cultures? Not in my U.S. ROYGBIV experience, but perhaps elsewhere, pending the results of an exhaustive search I have moved it out of the spectrum table and placed it here for safe-keeping -- FretPorpTine

I think it should go on the table since it is definitely a pure spectral color. The more we list, the better; I don't see how it helps to omit some pure spectral colors. The Encyclopedia Britannica lists it, but omits indigo. --AxelBoldt

Why were the colors made subpages of the color page? I don't see any good reason for this at all. I'm going to change it if someone doesn't convince me not to. (Actually, I hope someone else will change it.) --LMS

That colored text on the previous page hurts my eyes. Just wondering if the following table (or some variant of it) would be better or worse. --KQ


 * I like it a lot better and will steal it right away :-) --AxelBoldt

You can't steal what someone gives freely. :-) --KQ

I like the table, but I think using numeric values rather than color names would be more consistent across browsers ("green", for example, is a deep forest green on my browser, not a pure rainbow green), and I'd make it a little clearer that the English color names are only approximations, not actual distinct bands (though our brains do chop it up for us perceptually sometimes--but that's another article). I'd do it this way (with colors at 30&deg; hue intervals):

--LDC

Violent and Magenta aren't really quite the same thing - it's a minor difference (like between indigo and blue), but magenta is not a spectral color.

I agree; this article is "color" not "electromagnetic spectrum". To me, that means it should focus on the human subjective experience of color. That's why I include all 360 degrees of the color wheel, and explicitly mention that the purples are mixtures rather than pure spectral colors. --LDC


 * But the table is about the rainbow spectrum of chromatic colors; maybe we should mention the color wheel and HSV space separately --Axel

I think that's an improvement, Lee. But it seems less than a 30&deg; interval in the greens; the differences between green and either blue-green or yellow-green seems very subtle compared with the difference between any two other neighboring colors. I wonder if that's a problem with browsers on winME (I've tried Netscape 6.1, IE 5.5, and Opera 5.0) or with human perception. Does anyone know?

Oh, and in the top table Indigo does not show up in Opera 5.0. I guess they don't recognize that tag, which IIRC should mean it's not technically W3C compliant. --KQ

This one's better, but the wavelenghts seem to be a bit different from the ones we have now. I don't know which ones are right of course; I got the current ones from EB. --Axel

The "correctness" of the wavelengths will also vary with browser/video settings, so I might leave them out entirely (or put them into an article about the electromagnetic spectrum). My 11 colors are 30&deg; intervals in HSV space converted to RGB; a different conversion might yield slightly different results. Until we get more info in here (like a CIE Lab chart), I think it's sufficient. --LDC

I'd prefer the (correct) wavelengths to be there. After all, how else can you define "blue" if not by giving its wavelength? --Axel

You misunderstand my point--the wavelengths are correct, for some monitor with certain settings on some video card on some coomputer with some software. It is not possible for it to be correct on all of them, because HTML doesn't specify colors as wavelengths--as well it shouldn't, because that's not how human eyes perceive it anyway. --LDC I'm not so concerned about the HTML colors -- of course they will be off on most machines. But I want the right color names for the right wavelengths. So if we call it cyan, it should be listed with cyan's wavelength and some vaguely cyan-like HTML RGB mixture. Also, the table occurs in the physics section of the article, and spectral colors are cleanly defined physically by their wavelengths. --Axel

The article is actually wrong when it describes how color is percieved. It is right about the human eye having three types of cones, but it is wrong about what wavelengths these cones are sensitive to. The "Red" cone doesn't correspond to the color red, in fact its sensitivity is at 560 nm (somwhere around where green is in the charts given). "Green" is sensitive to 530nm, and "blue" is sensitive to 430nm. These are peak sensitivities, of course, since the sensitivities are skewed bell curves ("Red" and "Green" have tails that extend in to the shorter wavelengths, and "Blue" has a tail that extends in to the longer wavelengths. They all of other tails, but they aren't as long).


 * I don't understand why you say the article is wrong. The three wavelengths you mention above are given in the article; the overlap of the sensitivity curves is mentioned as well. I don't see how the fact that the cone's outputs are already preprocessed in the eye does affect the three-dimensionality of the color space. It should definitely be written up in an article on color vision though. --AxelBoldt


 * The problem with the article is that it characterizes color vision as positive mixing of the signal inputs. Signal preprocessing in the eye means that some of the inputs are subtractive as well. This is why we don't percieve only three colors and mixes of them. Yellow, for instance, is distict from both blue and green even though, chomatically, they it is a mixture of the two. Also worth noting is that I have never seen indigo in a color spectrum. My source isn't the most reliable (high school physics teacher), but he claimed, "The spectrum isn't ROY G BIV, it is actually ROY G BV. People added indigo because they thought that the number seven was a special number, i.e. seven orifices on your head." So that the color space, as detected by our eyes, can be represented as a linear combination of red green and blue (although our eyes are actually sensitive to a full spectrum). Perceptually, in our minds, the color space has, potentially, 6 distinct colors.


 * Perhaps it would be nice if someone would take a good picture of an actual spectrum and label it instead of using internet tags.--BlackGriffen

Color perception isn't a binary operation where if a cone detects a color it sends a positive signal to the brain. In fact, image processing starts right in the eye itself. Rods and cones are divided in to roughly circular receptive fields. Each field represents input to bipolar cell, the type of neuron that carries visual signals back to the brain. I don't know whether or not the fields can overlap. In the center of each field is a circle whose diameter is about half the diameter of the field. The cells in the center of the field are either a type of cone or a rod. The outer ring of the field consists of rods if the inner field has rods, or cones that may not be of the same type if the inner fields have cones (in fact, I believe that the cones must be of a different type, since rods detect brightness). The inner field of cells inputs directly to the bipolar cell, and stimulating only that field increases the number of action potentials to the brain (n.b. there is a base line action potential frequency in the absence of light). The outer field inputs to a cell called the horizontal cell. The horizontal cell takes that stimulus and sends an inhibitory signal to the bipolar cell. This means that if only the outer field is stimulated the input to the brain can go down below baseline. If the entire field is stimulated, the exitatory signal will be greater than the inhibitory one because the number of cells giving the bipolar cell exitatory input outnumber the single cell giving the bipolar cell inhibitory input. Their numbers even outweigh the fact that the horizontal cell inputs closer to the axon hillock (still above the cell body, though).

Ok, so we have a physical description of the "pixel" of the eye, we're almost done explaining color. The trick is to realize that the surround will actually subtract its sensitivity from the sensitivity of the center to create the sensitivity profile of the field. So, when "Green" cones surround "Red" cones, subtraction of their sensitivity curves (with possible weight factors inherent in the physical design of the eye) yields a field with a peak sensitivity at 650 nm. Just by changing the arrangement of the cones ("green" center, "red" surround) yields a field with a peak sensitivity to about 500nm, a genuinely green field. Using this model, it is easy to demonstrate that the eye is sensitive to 6 different colors (9 if it is possible to have the same types of cones in center and surround).

Also note that this shows that some optical illusions are possible because of a design flaw in the eye itself: this field arrangement exaggerates high contrast boundaries in the signal to the brain. --BlackGriffen

I wouldn't call that a design "flaw"--that's just a built-in edge-detection filter, that works much the same way a digital convolution would. Edge and motion information is generally more important to survival than color, so it's natural that we would have evolved to favor it. In other words, that's not a bug, it's a feature. :-) --LDC


 * If we plot the spectral colors on a chart showing the corresponding stimuli of the receptors, we find they define a curve, and that only colors inside the curve can be observed. Also, given lights corresponding to two or more points, we can duplicate all the points (up to human perception) in the region of the chart they bound by mixing them.

I think we need a picture here, maybe DrBob can help? The way the chart is set up is not clear to me. Are the colors modeled as points in three dimensional space, the three coordinates of every color point being the responses of the three receptors? --AxelBoldt

I set up a primary colors and primary pigments page so that a more detailed (or specific perhaps) treatment can be given to both. Really, so that eventually I could isolate printing and painting issues from science issues. There is some duplicated information in them as of right now.

Also as this can resolve the "are black and white colors" since in light white is, and in pigment black is.

--Alan D

The recent changes added a mentioning of a "Zone theory" involving six primary colors. If anybody knows anything about this, it should be added, or a separate article written. I am highly skeptical though: if the input from the eyes is 3-dimensional, then no matter what trickery the brain does, it will never create a six-dimensional signal out of it (a lower-dimensional signal would be possible, but pointless). AxelBoldt, Saturday, May 25, 2002