Wikipedia talk:WikiProject Color/Normalized Color Coordinates

As any formally trained color scientist would confirm, it is inappropriate to use unqualified color coordinates to specify a color appearance. In the color item articles, as I would guess all the other color-related articles, there is a box on the upper-right of the page that pretends to state the color coordinates of the color in RGB, CMYK, HSV and Hex values. This is wholly incorrect. RGB values in themselves cannot depict color appearance. In order to do so, the RGB triplet has to be connected to physical reality through, at the minimum, the definition of the RGB color space used, the white point of the illuminant, and in certain cases, the gamma function. The ICC specifications can help one understand what these requirements are. If Prussian Blue has such and such color coordinates, they have to be qualified as, for example, in AdobeRGB color space. Better still, linear XYZ color coordinates should be used. Furthermore the conversion of RGB to CMYK is wholly dependent on the particular transfer functions used for a specific set of inks and paper combination - the idea of posting the CMYK color coordinates for a physical color appearance is naïve at best and mostly preposterous.

As such the article runs contrary to the most basic precepts of color science, and is wholly misleading to uninformed readers. Trying to digitally reproduce the appearance of Prussian Blue using the information contained herein will not work at all.

Use of sRGB as a connection space neutral ground for RGB triplets is laughable. When specifying colors that relate to the physical world, there are radiometric and photometric values one can use, that have been designed specifically for that. Photometric values such as those standardized by the CIE standards body are the ones that should be used. They won't be bound to the specific gamut of display color spaces such as sRGB. Many natural pigments fall completely outside the gamut of sRGB, and henceforth, your entire SRGB-normalized reference system collapses like a deck of cards. Being aware of the CIECAM02 model of color appearance would help convert the CIE coordinates into practical RGB values for users who wish to simulate naturally-occuring colors under specific constraints of illumination, and avoid some surprises with metameric behavior.

As it currently is, the system used in Wikipedia is completely inadequate.

There is plenty of reference available on the subject, from books by Billmeyer & Salzman for the less technically enclined, to more recent works. An authoritative source on the topic is definitely Prof. Mark Fairchild of the Munsell Color Science Laboratory in Rochester, NY. — Preceding unsigned comment added by Ppanzini (talk • contribs) , 2006

Web color is only very recently starting to be aware of the whole "sRGB not enough" idea. The only two useful things in the infoboxes are really sRGB (because it's the web and we got to display something, duh) and CIELCH(uv) [hey it does have a one-to-one mapping to XYZ]. --Artoria2e5 🌉 06:10, 28 March 2023 (UTC)[reply]

I've clicked on the color infobox's a few times to end up here, and I'm not sure that this page is entirely appropriate as a link from mainspace articles - the presentation is geared very much towards colour value syntax conversion and says little about the infobox or values contained therein. Would anyone agree that we need a colour analogue of Wikipedia:How to read a taxobox ? Nihiltres^(t.c.s) 16:27, 8 May 2007 (UTC)[reply]

I've started a draft of a page at Wikipedia:How to read a color infobox, though it needs major work. I hope you don't mind that I used the American spelling, but figured it should match the spelling used by the infobox. 03:54, 9 May 2007 (UTC)

In the section on HSV coordinates it states that a fully saturated color should be at 50% value as in the Munsell system, but the example box shows a fully saturated orange with a value at 100%. Which is the preferred method for Wikipedia? Justin Foote (talk) 00:38, 28 September 2008 (UTC)[reply]

I agree with Ppanzini's remark above. Though RGB is useful for talking about colors on monitors, it is not a complete, standardized color system. Why aren't we showing CIE and NCS coordinates, which are complete and standard. --Macrakis (talk) 18:32, 20 January 2020 (UTC)[reply]

I agree, from a color science point of view, standardized color coordinates would best as the reference, with RGB, etc. conversions from the reference. But to do that, we need authoritative references for colors in terms of those standard coordinates. Instead, we have sources for Prussian Blue, as for instance Colorhexa and handprint. For colorhexa, the RGB coords are the primary key and CIE just a conversion with fake accuracy. Handprint gives the pigment PB27, and "average CIECAM J,a,b values for iron [prussian] blue (PB27) are: 19, -17, -36, with chroma of 40 (estimated hue purity of 45) and a hue angle of 245." The average across paints is great data, but handprint is a primary source and I don't know that it is considered a standard. If there is an authoritative source that is widely accepted as a standard across color-using industries and is accessible to us, let's use it. The Colour index is a standard in the textile industry, but I don't know that it extends beyond textiles. Thoughts? --{{u|Mark viking}} {Talk} 19:26, 20 January 2020 (UTC)[reply]

I know this article isn't a typical Wiki entry, but considering its explanatory purpose, shouldn't it still be held to the same content standards regarding the interjection of clear opinion and unverified claims? For example, you have this passage:

CMYK is generally something of a disaster, except as used in the internals of a printer driver. Unlike RGB, which is well-standardized as sRGB and which generally has a 1:1 mapping between coordinates and colors, CMYK is device-specific.

I won't claim to be the most learned on this aspect, but as I understand it, this is incorrect. In my multiple dealings with the print industry, color layouts are always requested to be submitted in CMYK both because of CMYK's palette consistency as well as the inherent shift that occurs when converting RGB colors to CMYK. I won't say that CMYK is perfect or better than any other color system, but to refer to a system which is the lifeblood of an entire industry as "something of a disaster" and as "device-specific" (which would mean that, on top of multiple print houses houses having varying color outputs on a given layout, most print houses with more than a single printer would *also* experience this phenomenon) seems a bit disingenuous at best. 107.77.204.154 (talk) 22:58, 15 November 2021 (UTC)[reply]

It is very device-specific in the same way "RGB" is. When you say CMYK there are unlimited ways to define what each ink-paper combination is, just as with RGB there's sRGB, Adobe RGB, P3, etc. The closest thing to a "universal" CMYK electronically is probably SWOP since almost every computer comes with one of their profiles, but numeric color charts don't always care for it. Artoria2e5 🌉 05:47, 28 March 2023 (UTC)[reply]

It seems like the only way to specify a color in such a way that no matter how the color is produced, it should look, behave and have the exact same physicals attributes, is to specify the monochromatic parts of the electro magnetic waves that make up the color. A monochromatic electromagnetic wave can be characterized by its frequency or wavelength, its peak amplitude, its phase relative to some reference phase, its direction of propagation, and its polarization.^[1] The whole purpose here must be to be able to use the name of a color in a meaningful way. If there are different ways of interpret "red", then it can not really be used for anything but poetry.— Preceding unsigned comment added by 195.67.91.199 (talk • contribs) (2022 Apr)

We are humans. We do Metamerism (color). --Artoria2e5 🌉 06:13, 28 March 2023 (UTC)[reply]

@RekishiEJ: My understand is that HWB does not add much more to the existing HSV/sRGB situation, while the newfangled RGBs are cool but there are just too many of them. This leaves us with CIELAB and CIEXYZ, which aren't going to change anytime soon since they cover everything a human can see. And of these CIELAB is probably more intuitive, so we should go with just that.

But wait! CSS Color Module 4 quite heavily promotes Oklab, which does improve on things like these weird blue-purple hues. The important point for computer people is that it does not add computational complexity like the CIE CAMs do, so they are going for it to be the next thing. Wikipedia doesn't really care about that (and might actually worry about whether it's a fad), but being able to stuff in a CSS color is nice. Do we choose it over CIELAB instead? Artoria2e5 🌉 05:56, 28 March 2023 (UTC)[reply]

Ah welp, we do have CIELCh(uv). Why are we going back to LAB again? --Artoria2e5 🌉 06:11, 28 March 2023 (UTC)[reply]