Talk:Munsell color system

From Wikipedia, the free encyclopedia

	This article is supported by the Color WikiProject, a project that provides a central approach to Color-related subjects on Wikipedia. Help us improve articles to good and 1.0 standards; visit the wikiproject page for more details.
B	This article has been rated as B-Class on the quality scale.
High	This article has been rated as high-importance on the importance scale.
Article Grading: The article has been rated for quality and/or importance but has no comments yet. If appropriate, please review the article and then leave comments here to identify the strengths and weaknesses of the article and what work it will need.

Contents 1 Any modern munsell-based tools? 2 what does it mean that munsell is "deficient by modern standards" 2.1 deficiency 2 2.2 deficiency 3 2.3 deficiency 4 3 Diagrams 4 Terminology

[edit] Any modern munsell-based tools?

Are there any color pickers in Mac OS X (or elsewhere) that are based on the Munsell system? It seems that most tools are based on a hue wheel which changes luminosity as it goes around, making them much less useful for picking equiluminous color schemes. --jacobolus (t) 23:23, 26 October 2006 (UTC)

There's a Windows Munsell Conversion program from here. I've also found some web-based color pickers, but can't track down my reference for them. Argyriou (talk) 23:37, 26 October 2006 (UTC)

There's also EasyRGB's "Color Harmonies", which tries to pick colors which go together. Argyriou (talk) 22:54, 1 November 2006 (UTC)

[edit] what does it mean that munsell is "deficient by modern standards"

This gets a citation, but no explanation, so I'm tempted to take it out of the article, as it doesn't really inform the reader, and just sounds like trash talk. I have no problem with including such statements, but they need to also explain in what way it is deficient, etc. --jacobolus (t) 01:23, 19 February 2007 (UTC)

But the statement is explained, I cite: "it didn't have the correct hues for cyan and magenta and it didn't recognize these hues as local value peaks so it rendered the color circle of saturated colors as a linear gradient from light to dark".--MWAK 06:32, 28 March 2007 (UTC)

Alright, but I have no idea what that means. What is the "color circle of saturated colors"? That term is never defined, and as far as I can tell, the Munsell system doesn't have any such circle. So it needs explaining. As for cyan and magenta, could we perhaps have a picture or description of what the "correct" hues are, and why these are wrong? The criticism given is *very* harsh (systematic errors, empirical data that it's bad, etc., with no citations for that), it doesn't really explain what exactly that means, and it doesn't give any examples of a better system. Also, the cited sources don't really say the same thing as the criticism given in this article. Overall, the paragraph sounds very unprofessional IMO. E.g. what in the world are "modern standards", and what systems live up to them? It would be good if a color theory expert could clearly explain the deficiencies for us. Otherwise, as I said before, I'm tempted to cut those criticisms out, or at least cut out parts that aren't supported by the citations given. --jacobolus (t) 17:50, 14 April 2007 (UTC)

What I can find of those references does not appear to support the strong statements made. The Gerritsen, in particular, appears to be an article advancing his own color theory, while for the other reference, the abstract appears to claim the opposite of what this article claims it says. There probably are valid criticisms to make of the Munsell system, but these references don't seem to be it. Αργυριου (talk) 20:42, 14 April 2007 (UTC)

Yeah, that was my impression as well. That said, there do seem to be some criticisms of the Munsell system's perceptual uniformity (and we have systems like CIE L*a*b*, L*u*v*, and CIECAM as a result). But I am not familiar enough with the field, or these criticisms, to accurately summarize them in this article. So it would be nice to have someone familiar with recent research give a more objective summary. In general, it would be nice to get some more pictures showing Munsell's "color atlas" in three-dimensions, so that readers can understand it. I can try to figure something out on that front. It would also be nice to have a better explanation of color systems before and after Munsell, to show its influence on the field. --jacobolus (t) 21:08, 15 April 2007 (UTC)

[edit] deficiency 2

I just tracked down and read the Kuehni (2000) article, and it doesn't say anything particularly relevant to an encyclopedia article about the Munsell system, and it certainly doesn't imply that the system is somehow irrevocably flawed: “The comparison of REN, RERE, and UCS shows thalargely the same group of researchers has struggled over decades to determine the correct chroma scales… This is an apparent indicator of the difficulty in visually scaling differences equally in a quadrant…As a result, we have a degree of visual uncertainty both in the chroma and the hue scales.” is hardly the condemnation implied by the citation. --jacobolus (t) 10:50, 16 April 2007 (UTC)

Well, first of all we must differentiate between the earlier forms of the Munsell system and improved versions. Until the eighties the system certainly was deficient, simply because it was based on the incorrect assumption intensity was continuously increasing from blue to yellow. So the "Blue-Green" was shown darker than Green, although in fact cyan is lighter than green; and "Red-Purple" was shown darker than Red , whereas in fact magenta is lighter than red. But of course this is only true if that cyan or magenta is of an equal saturation with the colours compared — and in the earlier versions it simply wasn't. I can only advise you to read the books by Gerritsen. The Munsell Color Circle is implicit in its space: look at the colour space from above or below and then disregard all but the most saturated colours — et voilá it appears.

As regards the usefulness as a psychometric reference, I can only concur with the conclusions by the site you yourself linked to http://www.handprint.com/HP/WCL/color7.html#MUNSELL:

As we've seen, the Munsell system contains two significant problems: (1) a variety of discrepancies were found in the perceptual spacing of colors, depending on their location in the color space, and (2) the quantitative difference between colors could only be defined on a single color attribute (lightness, chroma or hue) at a time. These problems were not resolved in the Munsell renotation, and the Munsell system remained an inconvenient and uncertain basis for quantifying perceived color differences on all three colormaking attributes.

Of course for simple matching the system is sufficient. The dangers lie in scientific research. To give but an example: Berlin used the Munsell system in the sixties for his Basic Color Terms. Now it matters a lot whether test subjects see this:

Magenta
Cyan

Or this:

Magenta
Cyan

Probably nobody will feel inclined to connect the latter two nuances with any basic colour term.

I agree my Kuehni reference was problematic. I feel what I stated is logically implied by his work, but for an encyclopedia more direct statements are needed. --MWAK 06:33, 22 April 2007 (UTC)

I'm really confused. The whole point of Munsell's system is that it doesn't assign any colors to arbitrary names, but instead specifies perceptually uniform hues, values, and chromas based on numerical values. That is, instead of specifying "color terms", its goal is to move us away from a need for color terms, at all. In other words, 5P is not meant to be "Magenta", and 5BG is not meant to be "Cyan". As far as I can tell, hues used by the latest internationally specified color model (CIECAM02) are spaced quite similarly to Munsell hues. Your explanation remains unsatisfactory.

It's true that there are issues with the Munsell system, but they aren't (from anything I've read) at all similar to the claims you put in the article. And regardless, the criticisms must be clearly explained, and well documented to go in Wikipedia. --jacobolus (t) 07:22, 22 April 2007 (UTC)

Also note that human vision has been found to be non-Euclidian, so it is not possible for a color system to be completely perceptually uniform in 3 dimensions at once. Which is probably one reason the Munsell system, which does not attempt to define a three-dimensional distance metric, is still used, despite systems like CIELAB and CIELUV which do. You should read [1], and perhaps even scroll up and read that whole page, if you want more details. --jacobolus (t) 07:27, 22 April 2007 (UTC)

[edit] deficiency 3

I didn't make myself sufficiently clear; my example was about the old system version; its lack of saturated magenta en cyan made any linguistic research carried out with it flawed. Remember that there are still many researchers working with the old stuff. I know a few of them myself. You might have a point that the Munsell system, in that it doesn't attempt the impossible, might be called superior :o). But the real reason it is still used is, I assume, simply that too much had been invested in it already.--MWAK 13:27, 22 April 2007 (UTC)

It's true. You still haven't made yourself sufficiently clear. You made some reference to "before the 80s", which makes no sense, as the Munsell Book of Color has not changed since the 1940s. I'm really not sure what you mean about lack of saturated cyan or magenta; any color can be fit somewhere in the Munsell color solid. Perhaps cyan or magenta appears brightest at some non-integer Munsell hue? That's the only interpretation I can come up with for your statement. But that still doesn't make much sense to me. Maybe you can give a quotation about the cyan or magenta from a color scientist which would better explain? I also don't understand what you mean about the "incorrect assumption intensity was continuously increasing from blue to yellow." Given that blue and yellow are hues with a wide variety of chromas and values, I can't parse that statement. --jacobolus (t) 13:40, 22 April 2007 (UTC)

When you say "You might have a point that the Munsell system, in that it doesn't attempt the impossible, might be called superior..." I think you are missing that the creation of a color space necessarily involves trade-offs. The Munsell system, as far as I can tell, had until very recently (and maybe still has) the most uniformly spaced colors along independent hue, value, and chroma axes of any color model in widespread use. It never claimed to be perceptually uniform in all dimensions, so you are probably right that other models (like CIELAB) would be better for psychological research into color distance in diagonal directions. But CIELAB has problems of its own, not present in Munsell. So it's a trade-off. --jacobolus (t) 13:51, 22 April 2007 (UTC)

Allow me to be more precise. Consider the original Munsell colour solid. First of all we must understand this colour space does not represent — by far — the whole of human colour perception; much was left out as the solid was limited by the range of available pigments at the time. The Munsell solid correctly shows that its most saturated colours, those with the highest shown chroma, are not equal in intensity or value: some are lighter or darker than others. The "value" relationship between its most saturated colours is however not shown in a correct way. The "rim" of the solid, the line of most saturated colours, constantly descends from yellow to green to cyan to blue and then constantly goes up again from blue to magenta to red to yellow. So its saturated green is lighter than its saturated cyan and its saturated red lighter than its saturated magenta. In reality however saturated cyan and magenta are lighter than a red or a green of equal saturation — obviously as being "light secondaries" they are lighter. So in a correct colour space of this kind the rim should go down from yellow to green, then up again from green to cyan and then again down from cyan to blue, making cyan a local value peak. Likewise for magenta. But it doesn't. As the solid doesn't, its most saturated cyan and magenta are too dark (and thus less saturated than the other colours in the rim) and it fails in its intention to show colours of equal chroma equidistant from the axis. Therefore it is deficient.

These deficiencies have been remedied in several ways. The Munsell system has in fact been in a constant process of improvement, showing on the one hand a comendable potential for adaptation — and on the other hand it was far from perfect in the beginning. The development was very intricate and I certainly don't know all the details. A simple method was at first too keep the original colours but "squeeze" the rim in the cyan and magenta sector. This way the chroma is correctly shown but it at once is obvious that the solid doesn't contain a cyan and a magenta as saturated as the other colours. A more sophisticated approach was too make the rim "blunter" allowing the colour space to expand to cover more of the perception range. In the nineties at last an equally saturated magenta was made available for scientific research.--MWAK 06:55, 23 April 2007 (UTC)

Okay, well, in discussing developments after the Munsell system, it would certainly be reasonable to explain this, with citations, and if possible give some diagrams which attempt to show readers more precisely what is meant, as the verbal description is still very imprecise. Such saturated colors are outside the gamut of sRGB, but something could probably be figured out.

Maybe we can also show a diagram of the changing shape of the Munsell solid between Munsell's original conception and the renotations. I'm not sure where such a diagram can be found though. I was under the impression that the colors in the Munsell book of color hadn't changed since the 1940s. Is there some explanation of the subsequent changes available somewhere?

Also, how was it that the 1940s measurements of human perception of value didn't pick up that the colors values were wildly different, as you're stating? It was my impression that Munsell value was pretty close to CIE L*/10, and fairly consistent across chromaticities. Is this wrong, or does L*a*b* also have the same systematic errors w.r.t. magenta/cyan hues? Some of what you're saying still doesn't make sense. What do you mean that an equally saturated magenta was "made available"?

Also, why is it that the Munsell renotation scales are still used for judging color systems w.r.t. (at least) hue and chroma spacing, if they are so inaccurate and deficient? Why is the CIECAM02 model so much closer to Munsell hue/chroma spacing than earlier models? I don't disbelieve you; it just doesn't completely agree with the reading I've done recently. But it's also possible that I am misreading.

Thanks for the clarification though. I think the way forward for this article is to emphasize the unique properties of the Munsell system at its conception, along with its tremendous influence on later systems, and then discuss any deficiencies in that context. --jacobolus (t) 06:57, 24 April 2007 (UTC)

What I said above was very basic, so it was easy for me to give this little analysis. The questions you ask are precisely those needed to be asked, but to answer them in the level of detail and exactness needed I would have to remember much of what I have forgotten and refresh my knowledge of the literature a bit — or really quite a lot to be honest :o). I'll try to find a good reference, but it could take a while.

Some are more easily answered though. The value problem was not that the measurements were wrong or contained systematic errors but that they didn't fit well with the external representation: the solid as represented, as part of a colour metatheory. What they could and probably should have done was to adopt a new solid and colour tree, more on the lines of Gerritsen's colour perception diagrams of the seventies. So they did pick it up; but instead of making a radical break they tweaked. This was partly due to the fact that Munsell already in 1915 was aware of these problems but, lacking exact data and rejecting simplistic rotated colour cube solutions like William Benson's, deliberately opted to conform to conventional notions about constant value change. At the time it was probably the most practical thing to do. Another consideration was no doubt that, having grown into an important commercial system, it was advantageous to maintain as much continuity as possible. Today there is of course a vast range of Munsell products for doing research and this range is no longer limited by its underlying colour model; so colours are available outside the solid, among them a saturated magenta.--MWAK 07:29, 25 April 2007 (UTC)

[edit] deficiency 4

Okay, so one thing I still don't quite understand then. Is the whole space that the solid fits in deficient? That is, if the boundaries are pushed outward to reflect new pigment colors, etc. that could not be made at the time, are things fixed? Or is there some other problem with the hue/value/chroma breakdown. And a related question, do CIELAB and CIECAM have these same problems? If not, how do their equal-value slices differ from Munsell's? --jacobolus (t) 09:02, 25 April 2007 (UTC)

Again good questions :o). The "deficiency" I talked about can indeed be described as a boundary problem and this would be fixed by expanding the solid or tree. It's good that you refer to pigments: pigment limitations were of course a very important factor. It was very facile of me to talk about the need to show a saturated magenta when in fact no permanent magenta was available. The remaining problems largely are a consequence of the fact that the system is primarily not empirical but a conceptual tool to understand colour space. It has these fixed five "principal colors" that are equally spaced. This is in two ways infelicitous. On the conceptual level it gives the dangerous suggestion to the user that these are somehow "unique hues". On the scientific level the equal spacing means you have to solve the contradiction between having saturated colours as local value peaks or lows at R, Y and G (and their contrasts BG, BP and RP) — which is a must at the conceptual level — and the aim of perceptional uniformity. You can't optimise both (using empirical data) when the spacing is kept equal. CIE 1976 L*a*b*, on the assumption it is empirically correct, can by definition not have either of these problems. For CIECAM this is true also but, being in essence nothing but a glorified colour pyramid, it is of course an imperfect conceptual model of colour space. Its equal value "slices" have a very complex topology. That's why CIE 1976 L*a*b* was created ;o).--MWAK 08:07, 27 April 2007 (UTC)

On the conceptual level it gives the dangerous suggestion to the user that these are somehow "unique hues". – why? I don't understand how it gives this suggestion.

On the scientific level the equal spacing means you have to solve the contradiction between having saturated colours as local value peaks [...] and the aim of perceptional uniformity [...] – Again, I don't understand. It was my impression that the only goal was perceptual uniformity, with no care whether the 5 principal hues have any specific properties. In my understanding, they are just arbitrary hues, and have nothing to do with "local value peaks" or anything else. That is, it was my impression that Munsell's whole idea was to move away from imbuing any particular hues or colors with special names, and instead use a numerical system.

For CIECAM this is true also but, being in essence nothing but a glorified colour pyramid [...] – Um, CIECAM is specified in hue/lightness/chroma coordinates, and its color solid looks roughly like Munsell's, though of course it has all sorts of complications for adaptation and contrast and different illumination, etc. But I don't know where you are getting the idea it is a "color pyramid". The most recent model, CIECAM02, is AFAIK the most empirically accurate color model around right now. --jacobolus (t) 03:57, 28 April 2007 (UTC)

[edit] Diagrams

Okay, I added a self-made picture (which took a bit longer than I thought it would). This article still needs lots of work though. --jacobolus (t) 02:29, 16 April 2007 (UTC)

[edit] Terminology

Perhaps the alternative terms for chroma (saturation) and value (brightness) should be mentioned here - as well as color spaces HSB and HSV (HSB redirects to HSV in Wiki). Maybe also a link to the HSV article. -- 85.22.11.133

The introductory paragraph says: “value (or lightness), and chroma (roughly saturation)”. Eventually it would be nice to explain the difference between "brightness" and Munsell value (lightness), and between "saturation" and Munsell chroma (colorfulness), because they are related but quite different concepts, and the Munsell versions are vastly more useful (being based on human perception) than the HSV/HSV versions. For the article as it exists now though (that is, without a section discussing the relation between Munsell and other color systems, which I would like to add eventually), I don't think such an addition is needed. --jacobolus (t) 23:08, 24 July 2007 (UTC)