Talk:Dominant wavelength

CIE calibration arbitraryness

On the CIE color coordinate space, a straight line drawn between a given color and the coordinates for standard white (1/3,1/3) can be extrapolated out ...

I have a feeling that the given definition is somewhat arbitrary. The sensitivity spectra of the three types of rods[sic] have a well-defined shape, but there is no absolute way to determine the ratios between the sensitivities at any particular wavelength. CIE 1936 uses the arbitrary definition that the surfaces on the three curves (with wavelength on the horizontal axis) are equal, which means that a flat spectrum results in X,Y=1/3,1/3. I, being a spectroscopist, am more used to spectra with frequency on the horizontal axis, which would result in a completely different diagram if the normalization were done according to that spectrum. One could also have normalized the curves to give equal responses for 6000 K sunlight, or in a way which makes the steps in wavelength along the border more constant—the 1936 CIE has too much emphasis on green colors. Each of the infinitely many ways to calibrate the color space would result in a different definition "dominant wavelength" and CIE is only one of them. Although, now that I'm thinking of it, the tristimulus space image:gamut_full.png would have the same cross-section, so maybe it is mainly the definition of the "white point" that should bother me.

Furthermore, the point 1/3,1/3 does not correspond to a white from a blackbody radiator,[sic] it is slightly of the "color temperature" curve. This is a less-important point since it makes sense to talk about a dominant wavelength compared to a flat spectrum that has no dominant wavelength. -- Hankwang 19:49, 20 Mar 2004 (UTC)

The article looks good to me, though it could very much benefit from some diagrams, especially for the paragraph discussing lines intersecting parts of the CIE space graph. Having a couple of frequency-component graphs might be good too, in order to illustrate the concept of "this color has many component frequencies; this other color has only one component frequency; because of the way dominant wavelength works, they look like the same color to humans." Bringing up another point—does this principle work for organisms that have better or worse color perception than humans? If so, how is it different for other organisms? And how does it relate to mechanical color perception (such as a digital camera or photo scanner)? Devices that, like humans, only sense color through stimulus of a few specifically-tuned receptors (like RGB)? -- Wapcaplet 01:28, 25 Mar 2004 (UTC)

I agree about the examples. This article, like many on Wikipedia is pointlessly over technical, some simple picture examples could be worth a thousand words.

An organism or machine with more photoreptors would be able to distinguish finer nuances in the spectral composition of a light beam. Where we infer yellow from a stimulation of our red and green receptors, something that actually had additional yellow receptors would percieve a difference between pure yellow versus red+green. That's my understanding anyway. If something only had red and blue receptors (no green) it shouldn't be able to distiguish between green and purple. I guess that explains how color blindness works. Lomacar (talk) 18:37, 24 January 2008 (UTC)

Explanation section

@Curran919, Jacobolus, and Koavf: I really think that this article has lost something by the removal of this section. It may have been unsourced and have problems with tone, but there is still valuable stuff in there that hasn't been retained. Try reading the article from the perspective of someone who knows nothing about colour theory – it's pretty much unintelligible. For a start, the understanding that natural colours rarely, if ever, consist of a single monochromatic wavelength is needed before we can even grasp why the concept of dominant wavelength is even needed or useful. Nowhere does our article say, as this source does, that this parameter is a more intuitive way of representing colour than colour coordinates.

Also lost is the (correct) claim that a colour can be resolved to a wavelength line only because humans have only three colour receptors. For species that have more than three receptors there will be many more colours that resolve to another line like the "purple line" and there will be multiple of these lines. For such species, colour perception can only be properly described in a multi-dimensional colour space. SpinningSpark 14:12, 23 September 2022 (UTC)

Perhaps a “background” section could explain some brief context, and link to color vision, trichromatic theory, colorimetry etc. for those interested in further detail. (Unfortunately Wikipedia doesn’t have any particularly clear articles explaining the basics of human color vision for laypeople. It would also perhaps be helpful to have an article at colormaking attributes or similar.) –jacobolus (t) 16:02, 23 September 2022 (UTC)

I am in agreement about the content in principle, but everything in Wikipedia that isn't common knowledge needs to be sourced. The article needs to be improved, but not by adding original research or unsourced claims. It's better to have nothing than to have material that is unsourced. ―Justin (koavf)❤T☮C☺M☯ 18:06, 23 September 2022 (UTC)

That is a misinterpretation of Wikipedia principles. Everything needs to be sourcable, it doesn't necessarily need to have an inline cite. That is entirely consistent with WP:V. In the early days of Wikipedia we had a much more relaxed attitude to sourcing so there is still a great deal of material out there that is perfectly fine, but is entirely unsourced. Yes, one can challenge unsourced material thus putting it outside WP:V, but rampaging through the encyclopaedia removing every unsourced statement is not constructive. I fundamentally disagree that it is better to have nothing. Our task here is to build the encyclopaedia and provide information. Removing material that we know or strongly suspect to be mistaken is a good thing. Removing material that we know or strongly suspect to be accurate (and sourcable) most certainly is not. I note that you have entirely failed to recognise that I provided a source for at least part of what I said. SpinningSpark 10:03, 24 September 2022 (UTC)

@Spinningspark I may be new to this, but I don't understand how you are interpreting WP:V. The lede excerpts "Even if you are sure something is true, it must be verifiable before you can add it" and "Any material that needs an inline citation but does not have one may be removed" seem to be pretty clear. Can you clarify? The prior section had neither inline citations nor blanket post-reference. I preserved what I could find references for (i.e. verifiable) and removed what I could not verify, especially that which I also considered incorrect.

If you have a verifiable claim that you'd like to add or reword, then I don't see why you can't re-add it. The addition of dominant wavelength being more intuitive than CIEXYZ may be a good one, but in re-reading the deleted section, I don't even see this - or a similar - claim, so I don't know why it is part of your criticism of the deletion. In any case, I agree with the claim that dominant wavelength is more intuitive than CIEXYZ, but that it is more intuitive than any other coordinate system (HSV/HSL/RGB) is highly contentious. Curran919 (talk) 10:54, 24 September 2022 (UTC)

You write, "Everything needs to be sourcable, it doesn't necessarily need to have an inline cite." but that page reads in part: "All content must be verifiable. The burden to demonstrate verifiability lies with the editor who adds or restores material, and it is satisfied by providing an inline citation to a reliable source that directly supports the contribution." What is material (other than common knowledge) that can be added without citations? ―Justin (koavf)❤T☮C☺M☯ 10:58, 24 September 2022 (UTC)

The term verifiable means capable of being verified. It does not mean that the passage currently has an inline cite. Providing an inline cite is the required response to a challenge, but there is no requirement that says you have to challenge uncited material. Just because you can delete something doesn't mean that you should. Besides, even in this era where the citation police have more power than the Iranian morality police, general references are still considered acceptable (no need to tell me this article doesn't have any general references, I know already). Having said that, I'm fine with Curran919 removing stuff they have actuall looked for verification and failed to find it. SpinningSpark 11:33, 24 September 2022 (UTC)

That doesn't answer my question, unless you are saying that all information on Wikipedia can be unsourced. Are you telling me that Jimmy Wales was wrong when he wrote: "I can NOT emphasize this enough. There seems to be a terrible bias among some editors that some sort of random speculative 'I heard it somewhere' pseudo information is to be tagged with a 'needs a cite' tag. Wrong. It should be removed, aggressively, unless it can be sourced. This is true of all information, but it is particularly true of negative information about living persons."? If information that has no source should be removed, then it shouldn't have been added in the first place, correct? ―Justin (koavf)❤T☮C☺M☯ 19:36, 24 September 2022 (UTC)

Please stop throwing strawman accusations at me. Of course I am not supporting keeping material that can't be sourced. I am railing agaainst uncited passages being removed wholesale regardless of how likely it would be to find sources for them. I think Wales agrees; he adds the caveat "...unless it can be sourced." He did not say, "...unless the original editor happens to still be watching and can be dragged back to provide a source." That qualification could just as easily be interpreted as "...unless the challenging editor can find a source." Yes, I know WP:V says it's not the responsibility of the deleting editor, but you don't have to delete in the first place and it's always more constructive to look for sources than delete potentially good information. SpinningSpark 08:56, 25 September 2022 (UTC)

I don't recall writing that you support keeping material that can't be sourced. If information that has no source should be removed, then it shouldn't have been added in the first place, correct? When do you think it's okay to add information that has no source to Wikipedia and isn't common knowledge? ―Justin (koavf)❤T☮C☺M☯ 10:14, 25 September 2022 (UTC)

When you wrote "...unless you are saying that all information on Wikipedia can be unsourced" pretty much implies that. Whether or not it should be removed depends on what you mean by "has no source". There is a huge difference between Wales' "'I heard it somewhere' pseudo information" and someone knowledgeable in the field writing that "X-rays have shorter wavelengths than UV" without an inline cite because it didn't occur to her that information that can be found in any basic textbook on the subject would need one. If you can't see or accept that distinction then there is no point responding to you any further. SpinningSpark 11:36, 25 September 2022 (UTC)

I wrote, "If information that has no source should be removed, then it shouldn't have been added in the first place, correct?" And you asked, " depends on what you mean by "has no source".": sources are provided by citations. I don't even know what you don't understand about the question. The example that you then gave was one that is common knowledge and is true by definition. As I have asked you several times already: "When do you think it's okay to add information that has no source to Wikipedia and isn't common knowledge?" ―Justin (koavf)❤T☮C☺M☯ 20:34, 25 September 2022 (UTC)

n-chromatiity

"it's pretty much unintelligible" -- should I interpret that with the typical disdainful connotation of the word? I don't even understand your characterization. "Dominant frequency" is a technical term - a technical measure - not a colloquial term. It should appeal to the layman about as much as the "niche" engineering articles you've shared on your user page, i.e. they do require some underlying knowledge of color theory resp. electrical engineering. Yet you also lobby for the inclusion of a concept on higher dimensionality color spaces that raises the complexity of the article by an order of magnitude? I don't get it.

Not to mention that I don't even agree with the "claim that a colour can be resolved to a wavelength line only because humans have only three colour receptors". In tetrachromatic vision, the geometric definition of the dominant wavelength in a 3-dimensional chromaticity diagram would be different, but only as a question of adding a dimension. The resulting 3-dimensional Helmholtz coordinates would still be primarily defined with by the dominant wavelength, i.e. the spectral hue to which the color is closest. Even adding an accurate exploration of higher dimension concepts in this article could be possible, but none of Chromaticity, Spectral color and Hue - which I consider conceptual 'parent' articles to Dominant wavelength - ever mention higher dimensional color spaces; so why this article? And where are any of us going to find a source for that anyway?

Furthermore, the proportionality of spectral vs. nonspectral colors in nature - or anywhere else - is irrelevant to the existence of dominant wavelength as a measure. Since nonspectral, polychromatic colors do exist, there will always be this definition, regardless of how common they are. Also, I didn't see in that source where dominant wavelength is related at all to the lack of pure colors in nature? Curran919 (talk) 21:04, 23 September 2022 (UTC)

I don't mean to be disdainful by the use of "unintelligible". Every article on Wikipedia should be written with the layman in mind, even technical articles (and since you are looking through articles I have written, I know I am setting myself up here as I am as guilty of not doing that as anyone). This is a general encyclopaedia, not a technical reference. So at the very least the lead should be understandable to a non-specialist who has come across the term and looked it up here. I think even for specialists, a statement of why we use this concept, rather than some other analysis, would be helpful to them. Writing only for people who already understand the material doesn't achieve much. Those that understand it will rely on their textbooks rather than Wikipedia and those that do not are not helped.

I agree that "dominant wavelength" can easily be defined for any number of colour receptors. My claim (and this possibly is OR, but since it was never in the passage of text being discussed its ORiness is not especially relevant) is that the concept of dominant wavelength becomes increasingly less useful for increasing numbers of receptor types. Do you agree that, as our article still claims in the Dominant wavelength § Complementary wavelength section, colors on the line of purples cannot be defined by wavelength?. That is, mixtures of R-B will not resolve to a wavelength (except for a small range that resolves to violet due to the second harmonic of the R receptor). If that is right, then a five receptor system (let's call it R, Y, G, B, UV) as possesed by many insects will have many more lines that will not resolve. For instance R-G mixtures will never be perceived as Y, any more than purple will ever be perceived as green to humans. I count six such mixture lines in this system: R-G, R-B, R-UV, Y-B, Y-UV, and G-UV. SpinningSpark 10:54, 24 September 2022 (UTC)

For a 3-primary system, you could say that ~1/3 of the triangle is undefinable (the so-called neutral-red-blue triangle). For those colors, dominant wavelength is not useful. For a 4-primary system R-G-B-UV, you need to introduce a third variable, so the helmholtz coordinates could be dominant wavelength, chroma1 and chroma2, or even dominant wavelength, secondary wavelength and chroma (its really speculative as to what kind of perception that extra dimension would evoke as it's highly psychophysical). The test color does not need to be linearly between the spectral locus and the neutral point as it does in the RGB system, because we have the extra variable.

If we break the tetrahedral chromaticity into 4 smaller tets which meet at the neutral point, then I think it is colors in the tets defined by non-sequential primaries (R-G-UV and R-B-UV) that would be undefinable, so ~1/2 are undefinable.

You could use this to say that dominant wavelength is a less useful definition in higher order space, but this is trivial. Clearly a single parameter is going to be less useful of a simplification (estimation?) of a subset of parameters the more parameters in that subset. Just like number of lines is a useful definition for polygons, but number of edges is not really a useful definition of polyhedrons. Curran919 (talk) 11:59, 24 September 2022 (UTC)

Given a 3+ dimensional chromaticity diagram "dominant wavelength" would be insufficient to describe direction, and a higher-dimensional analog would need 2+ parameters (anyone curious about this should hunt down papers describing color vision for various birds, insects, etc.). But this seems largely out of scope for this article. –jacobolus (t) 15:48, 24 September 2022 (UTC)

Agreed, it's out of scope to discuss higher dimensional systems when there is no clear use of dominant wavelength in sources for these cases. What is not out of scope is to say that dominant wavelength is only truly useful in 3-receptor systems. SpinningSpark 16:46, 24 September 2022 (UTC)

@Spinningspark if someone finds a reference that says its ONLY applicable in trichromatic systems, I'll cede the point. Otherwise, I think we'll be stuck with a viable compromise, something along the lines of "dominant wavelengths are a tool in trichromatic color spaces" that doesn't comment on other dimensions. Otherwise, speculating on higher order color spaces is not appropriate. Hell, as a dichromat myself, (something that would be much more appropriate to comment on), I don't think I'd be able to find a source that reflects my personal experience, and don't think it makes sense to include here anyway. Curran919 (talk) 17:25, 24 September 2022 (UTC)

I've been looking for a source that discusses dominant wavelength in higher systems and not turning up much. On the other hand, the fact that papers on the subject of 4- and 5-chromat species don't use the concept might in itself be saying something. I like your suggested compromise. SpinningSpark 09:16, 25 September 2022 (UTC)

@Jacobolus for every color in a dichromatic colorspace, there is a spectral color that gives an identical perception. So for dichromats like me, the dominant wavelength is a sufficient definition for 'fully' classifying color. This is unlike Trichromacy where this additional "fudge factor" purity is needed and only some colors can be mapped to it. So by saying that dominant wavelength is a good measure for tri- and not tetra- is very short sighted. You've already extrapolated it from dichromacy. Extrapolating it further to Tetrachromacy is mathematically the EXACT same treatment. You just need two fudge factors instead of one. Curran919 (talk) 09:49, 25 September 2022 (UTC)

The term “dominant wavelength” is used in practice to describe physical measurements rather than perceptual responses. In skimming a few papers about dichromatism, they don’t use the term “dominant wavelength” the way you are suggesting here, but rather use it as a technical description of their experiments, still with reference to a trichromatic colorimetric system. I don’t think it’s fair to say that a 2-dimensional chromaticity diagram in a trichromatic system consists of “dominant wavelength plus fudge factor”. Dominant/complementary wavelength as a concept exists because it is a physical quantity roughly correlated to perceptual hue, with colorimetric purity likewise roughly correlated with colorfulness [the psychological experiences of hue and colorfulness are substantially different from each-other]; from what I understand in dichromatic vision hue and colorfulness don’t really exist per se, being collapsed into one dimension (but I am no expert on the subject). In a tetrachromatic system the concept of hue is itself 2-dimensional. (Actually I’m not sure people have yet developed a good perceptual model for tetrachromatic vision, or have an idea how opponency works there, etc., for lack of human research subjects.) –jacobolus (t) 14:39, 25 September 2022 (UTC)

Human tetrachromats do exist, especially among females, but research has been limited. According to The Oxford Companion to Consciousness as many as 47% of female caucasians may possess the genes to express four cone pigments. It is not known how many of these individuals actually express those genes nor whether they process four colour channels to their visual cortex. SpinningSpark 17:53, 25 September 2022 (UTC)

I definitely see where you are coming from, but... I feel like we are getting into the realm of semantics here. The thing is, concepts that exist analogously in di-, tri-, tetrachromacy are going to be named to best reflect trichromacy, obviously! We dichromats are also so accustomed to living in a trichromatic society, that our own observations are tainted by the perceptions of trichromats. Sometimes I feel like I understand hue and saturation intuitively, despite clearly not experiencing them, but really I just understand them mathematically. Like you say, what I experience can be called a collapse of hue and saturation, but I think that's backwards. The more dimensions you've got, the more layers of abstraction you will have. A monochromat sees things in basic "LMS" space (or "S" space), i.e. the excitation levels of their one remaining cone. The second level of abstraction is opponent process channels, which is how I see my world, essentially as a single opponent channel. Then the trichromats - who have two chromatic opponent channels add on another layer of abstraction to differentiate into hue and chroma, because its more effective then thinking in two opponent channels. Tetrachromats would have to add another and we can't even speculate what that would be... To say they'd have an extra hue channel or extra chroma channel ignores the fact that di- and trichromacy both added levels of abstraction that the lower dimension individuals could not imagine. Anyway, with dominant wavelength, the fudginess is evoked by the name "dominant". In dichromacy, it would just be "equivalent wavelength", because the function of color = f(wavelength) is invertible, i.e. its a 1-to-1 relation... its exact. This is unlike trichromacy where multiple colors (chromaticities rather) map to the same dominant wavelength. But dominant/equivalent wavelength are really talking about the same thing... the 1st Helmholtzian coordinate. If trichromats were a minority in this world (well they are, but not in the human world), then this measure WOULD be called "equivalent wavelength" and you as a trichromat minority would probably use that name to describe the trichromatic 1st Helmholtzian coordinate. So is it correct to call the tetrachromatic 1st Helmholtzian coordinate "dominant wavelength"? Well... I would answer that in 5 different ways depending on context. Should this article reflect the first Helmholtzian coordinate for all dimensionalities? Or should it be exclusive to the usage in trichromacy? Should I make another article describing equivalent wavelength (assuming I can find a source)? Oh, and as to the existence of strong human tetrachromats... I am solidly on team negatory, but that's a discussion for Talk:Tetrachromacy. Curran919 (talk) 19:07, 25 September 2022 (UTC)

Tetrachromats would have to add another and we can't even speculate what that would be. – sure, hence my parenthetical qualification. –jacobolus (t) 20:31, 25 September 2022 (UTC)

because its more effective then thinking in two opponent channels – this isn’t quite why though. Hue and colorfulness are experienced as qualitatively different phenomena; two colors of the same "hue" seem to be qualitatively alike irrespective of colorfulness, whereas for two colors of the same "blueness–yellowness" but different "redness–greenness" (or vice versa), they just seem to be fundamentally different (based on a difference in hue) and it takes deliberate contra-intuitive training for e.g. an artist or designer to learn to recognize when two colors are the same on one of these color–opponent channels. (Which is not the same as for lightness, which can be easily distinguished). –jacobolus (t) 20:39, 25 September 2022 (UTC)

is it correct to call the tetrachromatic 1st Helmholtzian coordinate "dominant wavelength"? – I would personally expect that to not be the most sensible nomenclature. But it doesn’t too much matter what I think for the purpose of Wikipedia; in practice this is not something talked about in this way in published literature. –jacobolus (t) 20:43, 25 September 2022 (UTC)