User talk:Jacobolus/Archive 2013

Nomination of Unique hues for deletion

A discussion is taking place as to whether the article Unique hues is suitable for inclusion in Wikipedia according to Wikipedia's policies and guidelines or whether it should be deleted.

The article will be discussed at Wikipedia:Articles for deletion/Unique hues until a consensus is reached, and anyone is welcome to contribute to the discussion. The nomination will explain the policies and guidelines which are of concern. The discussion focuses on high-quality evidence and our policies and guidelines. Incnis Mrsi (talk) 09:30, 17 January 2013 (UTC)

A kitten for you!

I am very lucky to get your article about “hsl and hsv"。 it just solved my big problems about many conceptions. thanks. but i can't find how to derive that formula about hue??? the diagram is not so clear. if you use more clear geometry graph to demonstrate , i think it would be better.. my mail :　qduwg@163.com mr. wang. Qduwg (talk) 02:10, 30 January 2013 (UTC)

Mr.Jacobolus: I want get your mail for talking some problems about HSV/RGB, would you give me ? Thanks . best regards. My mail: qduwg@163.com Mr.wang Qduwg (talk) 11:41, 30 January 2013 (UTC)

Hi Mr. Wang. My email isn’t hard to find, it’s jacobolus AT gmail. –jacobolus (t) 19:34, 30 January 2013 (UTC)

… stuck in a 19th century mindset about how color vision works

Are you upset that I contested your original research on how to map Munsell hues to sRGB? I am possibly stuck in 19th century with my real projective spaces, straight lines, and convex sets, but I think that your “high color science” is far from infallibility. Open this colorimetric sheet, make fgrep ' 5B ' to extract relevant rows, locate them on CIE 1931 chromaticity diagram, and you’ll see that for a reasonable colorfulness 5B do not belong to sRGB at all. Of course, for desaturated variants the hue can shifted with various chromatic adaptations, encoding ambient white points, and so on beyond recognition. After I attracted attention to several of such high-color-science deceptions, your predictably become to make remarks about my mindset. Popular standards, such as sRGB, is a lie. Thanks to Wikipedia I realized how much they lie. And you certainly know how much does sRGB lie too. Incnis Mrsi (talk) 08:17, 22 June 2013 (UTC)

There’s no original research involved anywhere along here, but in any event, I’m not upset, just frustrated, because I feel like I’m talking to a brick wall. You are looking at the rows for hue = 5B at value = 1, which is something very close to black. That’s an entirely different color than we’re talking about. To usefully see what these Munsell colors look like on a computer display, you must (1) find a reasonable value/chroma at hue 5B, then do a chromatic adaptation from Standard Illuminant C (by which the Munsell colors are defined/measured) to Illuminant D65 (I recommend CAT02 or if you prefer the Bradford CAT), then convert from XYZ to linear RGB space, and finally apply the sRGB gamma function. Until you understand how those steps work, you’re going to be stuck. –jacobolus (t) 02:06, 23 June 2013 (UTC)

I did not say anything about V = 1, where for C = 10 the sheet lists a blatantly unphysical chromaticity (look at x, y values; http://www.cis.rit.edu/research/mcsl2/online/munsell.php apparently tries to explain this), but for C = 8 it does not differ greatly from, say, V = 5 at its maximal C: these are shades of approximately the same hue. I say you the third time: there is no reasonable [for sRGB] chroma at hue 5B: one has to choose between (approximation of) out-if-gamut colors and a use of tints, where your chromatic adaptations can distort the hue greatly. Incnis Mrsi (talk) 06:29, 23 June 2013 (UTC)

You are incorrect. Here’s a picture showing chromaticity coordinates for Munsell colors at value 5: http://i.imgur.com/vTtzlHo.jpg (note the contours shown are at steps of 2 chroma). I don’t have time right now to overlay the sRGB primaries on this picture, but you should be able to do that yourself if you desire. –jacobolus (t) 14:56, 23 June 2013 (UTC)

So what? I overlaid the triangle from File:Cie Chart with sRGB gamut by spigget.png and saw the same I talk to you for several days: 5B @ V = 5 runs along a slightly curved line (going from white approximately through steel blue, cerulean, towards spectral ≈ 481 nm), it crosses sRGB’s green–cyan–blue line shortly after C = 7 and continues towards the spectral locus until C = 16. V = 5, C = 7 is a tint (or a tone; it is not important). When you implement it as a paint under (a slightly bluish) Illuminant C and then switch to a slightly cyanish Illuminant D65, you expectedly push its hue further towards cyan. From there   was derived; I agree now that it is scientific, but it is unrepresentative. 5B is a fail for sRGB, which achieved a reasonable coverage of so named warm colors, which are traditionally preferred (BTW I read it at www.handprint.com, thanks :D), on the expense of these cool colors. Incnis Mrsi (talk) 16:40, 23 June 2013 (UTC)

BTW, to check the wisdom of your “chromatic adaptation from Standard Illuminant C to Illuminant D65”, could you compute sRGB values for Munsell white? ☺ Incnis Mrsi (talk) 17:09, 23 June 2013 (UTC)

That’s right. All of the colors along that curved line, when you paint a small patch with each one and put it on a neutral background, appear to have the same hue. All the bits outside the triangle made on the chromaticity diagram by the RGB primaries are outside the gamut of sRGB, so the brightest example we can see on a typical computer display is something like the 5B 5/7 that you mentioned (although as we’ve been discussing, that might be slightly shifted by the chromatic adaptation transformation). I don’t know what you mean by “unrepresentative”, “a fail”, etc. If you mean that sRGB is very limited in colorful middle-to-high-value blue and blue-green colors, that’s quite right. In response to your second question: these chromatic adaptations are defined such that they map white to white, so in this case Illuminant C will be mapped exactly to D65. –jacobolus (t) 01:48, 26 June 2013 (UTC)

One thing you might be misunderstanding here: 5B is not supposed to be any particular sort of “blue”. The Munsell “red”, “yellow”, “green”, “blue”, and “purple” hues were chosen to be equally spaced in hue, not to precisely match those color names. In fact, 5B is a bit on the green side of the “cyan” CMYK ink, and not within the range of colors that the ISCC–NBS system puts within the label “blue” (that would be 9B through 7PB), as you can see on that picture I pasted in the sRGB talk page discussion. –jacobolus (t) 03:58, 26 June 2013 (UTC)

Not 5B in general, but  . If you draw an (additive) chromaticity triangle for white, extreme (near-spectral) 5B, and process cyan  , then it will become process cyan more greenish than any color on the straight line white–481 nm. If your computation of the sRGB sample is correct, then apparently 5B in curved enough to envelop process cyan into its convex hull. Possibly, this is because Munsell’s “hue” corresponds to the dilution of dyes semantic, not one of mixing of lights. Incnis Mrsi (talk) 06:21, 26 June 2013 (UTC)

Photo request

Hi Jacobolus, I'm a user from the German Wikipedia so please excuse my lack of suitable English. At the moment I extend the German article about Edward Bernays which needs some photos to break up the gray. Here I saw that you take photo requests for Cambridge, Massachusetts – which is perfect. Bernays lived in 7 Lowell Street, Cambridge. You can see a picture of the Victorian house here. I want to ask if it is possible for you to take some photos of the house to illustrate the Bernays-articles in Wikipedia. I would be truly grateful. Best regards --Florean Fortescue 10:09, 9 August 2013 (UTC)

Ah, sorry. I should update that. I now live in San Francisco. –jacobolus (t) 21:05, 9 August 2013 (UTC)