Talk:Spectral color

Metallic color

What of metallic blues, yellows, brown and grayscale color? — Preceding unsigned comment added by 67.203.140.132 (talk) 02:47, 4 September 2015 (UTC)

physics of spectral color

i have added some material on the physics of spectral color and integration of the spectral colors in visible light. i have some further ideas on wavelength discussion, alternative light sources for visible light, etc, if a consensus indicates more such material would be desirable. i am open to editing or moving text i created just now to another article. im a newcomer to "color" editing so i am open to suggestion. Anlace 15:32, 20 May 2006 (UTC)

try

pls add the page non spectral colors. —Preceding unsigned comment added by 82.205.243.3 (talk • contribs) 20 June, 2007

… mixing a gray-scale color and a spectral color, such as pink or brown

Both are very inappropriate examples and quite ambiguous terms. "Pink" of course is not spectral, but it may denote not only a low saturation red (i.e. a mixture of red and white), but also those side of purple tones (≈350°) which is close to red (360°=0°). These colors obviously are not mixtures of gray(white) and red. Brown, as stated in article, has HSV: h=30°, s=100%, v=59%, which has maximal saturation, so is nearly a spectral color (orange), but with low intensity (dark). Incnis Mrsi (talk) 08:38, 17 April 2010 (UTC)

So, pink or electric blue?

There are plenty of sources on a cyan spectral color, usually mentioning such wavelengths as 502, 505, 520, sometimes 488 nm (note I don't claim that it is the same as the color of cyan ink). These are not only students' webpages such as [1][2][3][4], but also peer-reviewed publications [5][6][7], information about commercially available LEDs and lasers [8][9][10] and even a US patent [11]. So, do not do such edits[12] without a discussion. I hope, there are no doubt that the hue of electric blue should be called cyan rather than, say, blue or green.

On the opposite side, please, provide at least one source about "reddish spectral color" which gives an ordinary pink when tinted. Not just verbal declarations "RED + WHITE = PINK", but something demonstrating it numerically or on a plot. Without such a source, I will wipe the pink from this example without a mercy. Incnis Mrsi (talk) 07:40, 16 September 2010 (UTC)

I was coming fresh from the discussion at Visible_spectrum#Cyan_vs._Indigo. The use of "cyan" as the name of a spectral color is a very modern and still rather unusual variant. We should avoid it. I don't understand your point about pink, but find a better example; what is "electric blue" anyway? Dicklyon (talk) 05:08, 17 September 2010 (UTC)

Please, give your definition of pink, then we shall discuss my point. Some exact or approximate hue value, or some concrete interval of hues (of course I understand that hue is an angle and lies on the unit circle rather than on the real line). Incnis Mrsi (talk) 13:26, 21 September 2010 (UTC)

In it primary definition, pink is used to refer to colors of high lightness in the red or purplish-red hue range.   (a central – in terms of hue/lightness – representative for the ISCC–NBS category “pink”) or   (ISCC–NBS “purplish pink”) are reasonable examples. Pink is also often used for not-so-light purplish red and reddish purple colors (“hot pink”). –jacobolus (t) 21:15, 21 September 2010 (UTC)

The American Heritage dictionary has: “Any of a group of colors reddish in hue, of medium to high lightness, and of low to moderate saturation.” Random House has: “a color varying from light crimson to pale reddish purple.” Collins has: “any of a group of colours with a reddish hue that are of low to moderate saturation and can usually reflect or transmit a large amount of light; a pale reddish tint”. The name comes from various flower species in the genus Dianthus called pink. –jacobolus (t) 21:21, 21 September 2010 (UTC)

In any case, it’s not clear to me what it means to “tint” a “spectral color”. Tinting in general means adding white to a paint, and paint is by definition not spectral. –jacobolus (t) 21:23, 21 September 2010 (UTC)

Sorry for my absence for such long time. IMHO samples and definitions of pink mentioned above demonstrate that the pink color is not a good example of the sum of the white and a spectral color. I wrote the word "tint" in this discussion when mixing with the white light was assumed; I realise that tints used in painting are not exactly the same way of desaturation. If the current version of the article does not cause objections, let us stop here. Thank you for valuable comments. Incnis Mrsi (talk) 17:09, 10 October 2010 (UTC)

A spectral color is a color that can be evoked...

I think the article should use "can be evoked" instead of "is evoked" in its definition.

The human eye cannot distinguish between monochromatic light of a given frequency and a mixture of frequencies that evokes the same response in its (only 3 types of) colour receptors. In other words: they are the same colour. AlexFekken (talk) 10:23, 23 April 2012 (UTC)

Give an example of such "mixture of frequencies", and we will discuss this point then. Hint: each spectral color is an extreme point of the CIE chromaticity space. Incnis Mrsi (talk) 17:26, 23 April 2012 (UTC)

(Colour) television, computer screens, movies, photography... they all rely on this principle. Is that enough examples? AlexFekken (talk) 09:40, 25 April 2012 (UTC)

On what principle? AlexFekken, if the RGB's Green really "evokes the same response in your colour receptors" as one of spectral greens frequencies does, then your vision is probably deuteranomalous or even worse… I am sorry, but over 80% of humans have normal colour vision, for which RGB's gamut represents only a part of all chroma. Because spectral and near-spectral colours does not occur frequently in everyday life, colour distortions due to RGB are not strong on most images. Incnis Mrsi (talk) 11:28, 25 April 2012 (UTC)

So are you saying that when you have normal vision and you look at a scene which is dominated by a spectral colour (e.g. a wall coloured evenly with a single spectral colour) and then look at a "perfect" (i.e. as good as possible) television image or photograph of it, side-by-side if you like, you would almost always (i.e. for every spectral colour except possibly for some R, G or B used in the colour reproduction itself) be able to tell the difference? That sounds very implausible to me. But I admit I have never done the experiment. AlexFekken (talk) 10:56, 27 April 2012 (UTC)

And from a mathematical point of view I would also expect that it will (again "almost always") be possible to tune the R, G and B levels so that the response levels of the colour receptors to the mixture exactly match those produced by the spectral colour. AlexFekken (talk) 11:15, 27 April 2012 (UTC)

The RGB green (even in sRGB) is nearly indistinguishable from the pure spectral version of this color; if you can tell the difference between them, congratulations, but for most people (with an average color vision) it's hard. The CIE XYZ chromaticity diagram is not perceptually uniform. In the CIE L*a*b* color space (which is close to perceptual uniformity), the sRGB gamut doesn't seem so small compared to the human eye gamut (especially on its three primary colors). Most modern monitors have a wider gamut than sRGB, so I guess that you should start to change the idea (that, unfortunately, a lot of people have) that "When talking about color vision, the human eye is close to perfect, it's magnificent, and we will never be able to create a device that can create all the colors that it can perceive". No, we haven't yet (there's probably not enough people interested in that), but it's possible. 8-leaf clover (talk) 12:41, 30 March 2023 (UTC)

Very wide gamut 3-primary displays are problematic because they necessarily use narrow-spectrum primaries which cause a high degree of metameric variability between viewers, which makes it impossible to present consistent color relationships to viewers with slightly different cone sensitivities. sRGB is a actually a pretty good compromise. If you want a display with wider gamut, it would be significantly better to include a bigger variety of primaries (a display with, say, 8 primaries could be amazing), but display vendors aren’t really interested because it would be more expensive and wouldn’t as easily fit with the rest of the existing pipeline of hardware and software. Nowadays GPU-like processors are definitely capable of doing the required image processing on the fly, but it would require some research effort. –jacobolus (t) 15:45, 30 March 2023 (UTC)

The article Color triangle illustrates the sRGB color triangle, the gamut of colors that can be made by mixing typical CRT R, G, and B primaries, and shows what ALL spectral colors are outside that range, and by implication of their place on the chromaticity diagram, distinguishable by normal human vision. This idea that it is "almost always possible to tune the R, G and B levels so that the response levels of the colour receptors to the mixture exactly match those produced by the spectral colour" is a common misconception. Not so. Dicklyon (talk) 15:18, 27 April 2012 (UTC)

Apologies for wasting your time: my mistake. Your links helped me along: I am (apparently) talking about Metamerism (color) and my claim was basically that metamerism can occur with spectral colours as well as mixtures. The hint about the spectral colours being extreme points makes sense to me now. AlexFekken (talk) 09:20, 28 April 2012 (UTC)

So, a color is evoked. OK, then it is a mental construct. I don't know of any 'single wavelength' sources. Citation needed. Make that an authoritative reference please. Last I heard, all sources had varying doppler shifts, electric field environments, magnetic field environments, and gravitational field environments which makes the population of photons have at least some range of wavelengths. That is, not "single wavelength". Why not use "mono-chromatic" and if it's felt necessary to define that, then explain it means a narrow range of wavelength/frequency/energy. Finally, if a source has a color which is indistinguishable from a spectral color, the lead here claims the color would not be spectral unless the source was only emitting a narrow band of wavelengths. Hmmmm. So, in order to determine whether a color qualifies, we must measure the spectra of its source (since clearly the human eye isn't able to make that determination). And lets say X% of the light is narrow ranged, what is X? 80% of light? (by photon count? by energy?) 95%? 99%? Given that everything emits black body radiation, you can't limit the light input to 100.00000...% pure spectral light, so what is the allowable minimum X? And then there's defining the allowable range. 10 nanometers? 1? and it that the distribution's width at half-height? it's standard deviation? 3 sigma? I disagree that a color that is indistinguishable from a spectral color isn't a spectral color. As an aside, it's claimed that the human eye is capable of perceiving about 10 million colors, so how many of those 10 - 20 million are spectral? I certainly don't know, but I think it's worth mentioning.174.130.71.156 (talk) 19:03, 30 October 2022 (UTC)

Chromaticity of spectral colors

Does somebody have a good, scientific source about this? The question refers not only to one of 2012, but also to WP: Articles for deletion/Unique hues. Namely, it should confirm two things:

• Any spectral color is an extreme point of chromaticity;
• Any spectral color's hue (but of extreme red and extreme violet) can be obtained with interpolation between nearby (slightly higher and lower) frequencies.

Of course, if a source also considers some psychological aspects, it would be a bonus. Incnis Mrsi (talk) 07:55, 18 January 2013 (UTC)

… that is, created by the addition of grey

https://en.wikipedia.org/w/index.php?title=Spectral_color&diff=540523315

A shame on all us that this “clarification” can be visible for more than 12 hours. The color people, why do you promptly go to war when I change red to cyan or so, but sleep when an idiocy is injected to the article? Incnis Mrsi (talk) 06:47, 11 June 2013 (UTC)

Something about laser pointers?

The three cheapest kinds of laser pointers are 650 nm red, 532 nm green, and 405 nm violet. The cheap ones are nominally 5mW, although the green ones tend to be 2-3x overspec, i.e., 10-15mW. Red covers such a huge swath of the spectrum that a 650 nm laser pointer doesn't tell us much. 635 nm laser pointers (more expensive) also look red, just not as much when there's a side-by-side comparison. 670 nm, 808 nm, and 980 nm laser pointers also look red. 532 green, on the other hand, is widely acknowledged as having a very slight yellowish tinge. A more rare and expensive laser pointer in a color known as forest green at 520 nm has a bluish tinge. Another more expensive forest green laser is 515 nm, trends even more towards blue. Green traffic lights in the US are now 505 nm, previously 507 nm (the old 3M lamps). However, traffic lights, even with diodes, are much more broadband. Green traffic lights are intentionally skewed towards blue so that people with the most common sort of color blindness can see them. If you're not paying attention, they look green. However, if you look closely, you are suddenly amazed at how blue they are, almost turqoise. The standard for yellow or amber traffic lights in the US is 594 nm. They also have amber laser pointers (more than $500 for 1mW) at 593 nm. In bother cases, they seem to really look amber.

589 nm laser pointers are marketed as "yellow" but that's kind of tricky. Most people agree that true yellow is further down the scale, although they can't agree on whether it's 580 or 590.

The cheap 405s you get on eBay are actually (and consistently) 403.5. This kind of laser is often advertized as Blu-ray because it's the wavelength used in Blu-ray DVD. However, the color is actually far violet bordering on ultraviolet. Another sort of laser is advertized as "blue" or "445/447/450." These are actually two kinds of lasers so close that their wavelengths overlap. Although they often show up as blue in photos, people who have actually seen them say there's a definite tinge of violet that sometimes shows up in photos. 473 nm laser pointers are marketed as "true blue" although they look more like sky blue. True sky blue or azure may actually be closer to 478. Not that this is necessarily the color of the sky, just the name of a tertiary color, halfway (perceptually) between blue and additive cyan. (Printer or subtractive cyan is a completely different color).

I forgot to mention orange. Here we need to look at bench lasers. 612 nm is a kind of orange, but has a distinctive reddish tinge. 607 nm has a very, very faint reddish tinge. 604 nm looks orange enough that it has neither a perceptible reddish not yellowish tinge.

One last point: people who don't have a lot of experience with lasers can easily be food by the Bezold-Brucke shift, where hues shift with brightness. However, when you've played with or worked with a particular wavelength long enough, you eventually see it in all it's incarnations, at which point you get some idea of where the center of gravity is. Zyxwv99 (talk) 22:46, 30 March 2014 (UTC)

Can't Dichromats distinguish between spectral vs non-spectral colors?

The article says "One needs at least trichromatic color vision for there to be a distinction between spectral and non-spectral colours..."

That implies that dichromats can't tell the difference between a laser and a gaussian emission centered at the laser. If you have two cones, the ratio of the two cones will change, thus the system has information to discriminate between the two.

Or is it talking about the line of confusion and how the spectral locus becomes confused for other chromaticities along a line?

Posted June 16th, 2020 by Willingo North8000 (talk) 16:48, 2 March 2021 (UTC)

@North8000 Dichromat and CVD researcher here. Our chromaticity diagram is a line. all of the non-spectral color collapse into the same line as the spectral colors. There is no distinction. We have no clear distinction between wavelength and purity (hue resp. colorfulness). See simply the neutral points where red-green dichromats see spectral cyan or spectral pink (rose) as gray. Take a pure spectral color and move it away from these points and you see an apparent change of hue and increase in colorfulness at the same time towards unique spectral blue or yellow. Everything that I see as and associate with a pale, desaturated color, could simply be a pure bright rose, or bright turquoise. We use an insane amount of subconscious context clues to know the difference though. Curran919 (talk) 07:52, 19 September 2022 (UTC)

@Curran919: That wasn't my post; I just added a note on who posted it and when. Sincerely, North8000 (talk) 16:25, 19 September 2022 (UTC)

@North8000 sorry! Your name pops a lot! Curran919 (talk) 18:22, 19 September 2022 (UTC)

@Willingo: Then the above explanation is for you!

sRGB Sample methodology

Incnis Mrsi noted in Special:Diff/560409216 that the current "sample" displays are ad hoc and unprofessional. It doesn't need to be that bad: since the CIEXYZ color space of 1931 and the initial definition of the sRGB in terms of the XYZ, it has always been possible to take a wavelength λ and convert it to sRGB in a very standard way:

• First, turn the wavelength into CIEXYZ per CIE 1931 color space#Computing XYZ from spectral data, emissive case.
• Scale the radiance so that the signal is of about the right strength.
• Use the sRGB definition to convert the XYZ into sRGB.

The only knob to turn here is the radiance value. I recommend using a uniform radiance value so the differences in luminance is visible. The value should be chosen so that the fewest values run out of the sRGB gamut. For those that do, we will need to do chroma-reduction gamut mapping or something else.

As for the hue angle... I strongly recommended not using HSV/HSL. CAM16-UCS hue is a good fit of Munsell hue, and should be used here.

Someone (probably me) can do it in a Jupyter notebook so people can reuse the results. --Artoria2e5 🌉 15:50, 2 March 2021 (UTC)

Table Cleanup

The table of spectral colors is a bit of a disaster. The 'main' table was put in June 2013 and still reflects most of that uncertainty, having grown organically since then, lacking citations, lacking clarity, lacking cohesion. The vast majority of colors are not even spectral colors.

The lead table was put in November 2021 and I think incorporates all of the information that needs to be in the 'main' table. If we want a "list of colors of high purity", then that can be a thing, but I don't think it belongs in this article. I opt to scrap the 'main' table wholesale and leave the table in the lead. Will wait for some input before I do so. Curran919 (talk) 08:04, 19 September 2022 (UTC)

I have seen this before and thought that it's arbitrary and seems inaccurate as well. ―Justin (koavf)❤T☮C☺M☯ 20:04, 25 September 2022 (UTC)

I agree it's a mess. Take it out or reduce it to sourced items if you find any. The little table in the lead is also full of unsourced data. It might be better to have Newton's colors there (no cyan, yes indigo, I think is the difference); or some other sourced table. Dicklyon (talk) 21:04, 25 September 2022 (UTC)

Actually that's a good idea. I can make a table listing a few competing sources and keep each one to the basic 6-8 colors. Time to open my sandbox... Curran919 (talk) 07:23, 26 September 2022 (UTC)

Newton’s colors are dumb. But the name “cyan” is also out of place here: cyan is just jargon for the slightly greenish blue used in CMYK printing ("cyan" is the Greek word for "blue"), which is entirely different than what is shown here. –jacobolus (t) 14:58, 26 September 2022 (UTC)

I agree that Newton's colors are dumb, but they're also pretty classic. Is there some other set with similar stature? Just leave off indigo? Dicklyon (talk) 20:34, 26 September 2022 (UTC)

I think the psychophysical basis of indigo and cyan is well discussed and also well represented on wikipedia:

• Indigo#Classification_as_a_spectral_color
• Blue–green distinction in language
• Color term#Basic color terms
• ROYGBIV (not in there, but could be)

Do we want to also represent that discussion here? I was just going to pick a representative sample of sources and use those. So far I have samples with:

• ROYGBIV (2x with different cutoffs)
• ROYGBV
• ROYGCBV
• ROYG(light blue)(dark blue)V

Which I think at least frames the problem nicely. Then we can direct to whatever discussions on the merits of inclusion of indigo or cyan. However, I've never heard someone argue against cyan tbh. Most languages don't naturally separate their blues, but Russian, Italian and Hebrew do differentiate blue in their basic color terms, so I imagine there is a good psychophysical reason to do so. I'm just not sure if they are separated more on hue or lightness. Curran919 (talk) 08:36, 27 September 2022 (UTC)

Indigo is nice dye, but from anything I can tell the word "indigo" has never been popular as a basic color term (except in children’s mnemonics about rainbows derived from Newton’s list). The term “cyan” should really not be used outside the context of CMYK printing. In any other context is is somewhere between incredibly confusing/misleading and wrong. It was adopted by some people in the computing field in the 1980s–90s (e.g. in HTML/CSS colors) to refer to a color (a full-strength mix of their G (yellowish green) and B (purplish blue) primaries making a bright greenish blue) that had nothing whatsoever to do with the printing ink, which has caused a ton of confusion, but most English speakers do not regularly use the word “cyan” as a color name, and that was never historically a common term (except in printing). The wikipedia article cyan substantially consists of original research unsupportable by historical sources. If you want a more concise name for blue–green, I recommend picking “teal”. –jacobolus (t) 14:46, 27 September 2022 (UTC)

If you want a more source-supportable color list try ISCC–NBS system § Color categories. –jacobolus (t) 14:54, 27 September 2022 (UTC)

@Jacobolus yeah, ISCC-NBS is obviously one of the color term models included in my graphic. Even though it's technically a model to describe a full gamut, not just the spectral colors. Curran919 (talk) 18:32, 27 September 2022 (UTC)

@Jacobolus: I made a concept of the table: User:Curran919/colortable Curran919 (talk) 20:49, 27 September 2022 (UTC)

Contrast should be improved, and it may still be helpful to include sRGB approximations of the spectral colors. –LaundryPizza03 (dc̄) 09:43, 9 October 2022 (UTC)

sRGB equivalence would be pretty meaningless to approximate ranges of spectral colors. Not only because sRGB gamut is far from the spectrum, but also because you change all three values at the same time as you make your way down the spectrum. However, equivalence of hue degree (as exists in the current table) would be more meaningful and should be added. I need to find a reliable conversion source. Curran919 (talk) 12:32, 13 October 2022 (UTC)

What if we just copy something like the table from Visible spectrum#Spectral colors instead? (But perhaps without the little color boxes next to each name, and some clear indication that these categories are approximate and not precisely specified; I’d also skip the “cyan” category or rename to blue–green.) The color backgrounds in Curran919’s table are a bit hard to read, the specific colors used for backgrounds seem arbitrary and excessively visually heavy, and the inclusion of several different sets of break points for color names seems unnecessary. Including a diagram showing a gradient with labeled wavelengths is less unintentionally misleading. –jacobolus (t) 20:16, 30 October 2022 (UTC)

An even safer thing would be to draw a chromaticity diagram (I would recommend using the u'v' diagram), color in the triangle (gamut) between the sRGB primaries leaving the part part outside that area uncolored, and draw lines on the diagram corresponding to ISCC-NBS (or whatever) names, e.g. following Kelly (1943) "Color Designations for Lights". Then it is clear to readers that the colors drawn on their display are not actually "spectral colors" per se. –jacobolus (t) 20:28, 30 October 2022 (UTC)

@Jacobolus: That table WAS the lead image in the main article until this month. As the preceding discussion, the data therein is unsourced, I could not find a source that aligned with what it claimed. That's probably because it was edited after addition by another user; a theory made obvious by the fact that the wavelength/frequency/energy numbers don't actually match each other in some colors.

I won't defend anything about the visual presentation of the table, and especially not the color accents. I don't have an opinion on that in general (being colorblind and all).

However, I do feel more strongly about the inclusion of multiple disagreeing sources. It seems silly to include a singular source of cutoffs, then just below claim there is no authoritative source of cutoffs. Unfortunately, I also don't see a good way to aggregate many definitions of cutoffs into a simpler representation. Putting a range (uncertainty) on ranges (of colors) is going to be far more confusing, and putting in gradients or middle points is not going to be possible with the (sometimes large) disagreements. It's also important to demonstrate that different sources use different color terms and/or include or exclude colors (indigo/cyan).

I'd also generally be opposed to using a chromaticity diagram. This is an article about spectral colors only after all, not a dissection of all color terms. We also don't want to require someone to understand chromaticity diagrams before getting to the useful information. Do you mean something like this? I've also seen some color naming experiments where they present disagreement in color term cutoffs from a large population like in figure 2 of this paper, but I think it would be hard to adapt (and those aren't exactly spectral colors anyway).

In any case, I'd like to see a mockup of your suggestion. Curran919 (talk) 08:57, 31 October 2022 (UTC)