Talk:HSL and HSV/Archive 4

Archive 1

Archive 2

Archive 3

Archive 4

Archive 5

Archive 6

Comparison of HSL and HSV

This article used to have a section called "Comparison of HSL and HSV" that summarised the difference in a nice & succinct way, & was very useful to me, personally. I'm not sure if I'd have learned what I needed to learn if that subsection hadn't been there, so I'd just like to advocate putting it back. -anon —Preceding unsigned comment added by 124.171.82.96 (talk) 13:19, 8 March 2010 (UTC)

What part of it in particular did you find useful? Here's a link to it. It was filled with vague speculation/opinion ("some people ...", "... which may be considered ..."), and an inaccurate/misleading quotation from the CSS spec. The current version describes how each model is derived in great detail, and has an image at the top of the page which shows IMO quite clearly how each one works. But if there's some part that's unclear in the current article, we should definitely work to improve the explanation. Cheers. –jacobolus (t) 19:07, 8 March 2010 (UTC)

I agree with the anon. The article delves too quickly into obscure details. The article as it existed before you made your substantial changes was easier to understand at first glance. For instance, LCH is only indirectly related to the subject matter. SharkD  Talk  15:35, 10 March 2010 (UTC)

Okay. My contentions were/are (1) that the "easy to understand" part before was somewhere between too vague, misleading and inaccurate, and (2) the current lead section pretty much provides all the same information "at a glance" for those interested, including a couple of images at the top which should be sufficient for readers only interested in a cursory treatment. But is there some kind of section we could add near the top of this current version that would clarify things? What's your high-level vision for the structure of an ideal article about the subject? My imagined audiences are (a) people wanting a quick definition of HSL/HSV, who want to quickly skim and prefer looking at some pictures as a way to understand, and (b) readers looking for an in-depth explanation of why HSL/HSV are used and how they function, perhaps academic researchers, programmers working on an implementation, or artists or information designers who want to deeply understand their tools. For the former group, being sloppy doesn't matter so much, because they're not going to read especially carefully and won't mind if details are elided &c. For the latter group, providing careful and comprehensive context and description is essential. In other words, one person's obscure detail will save another person several hours and clear up a third person's long-held misconception. –jacobolus (t) 20:23, 10 March 2010 (UTC)

As for the relevance of what this article calls luma/chroma/hue: it is used in photoshop and 2-3 other image editors in exactly the same context that several more image editors use HSL or HSV (in photoshop, it is called "luminosity"/"saturation"/"hue", and in photoimpact "luminance"/"saturation"/"hue", names which are sometimes applied to the dimensions of HSL). There is widespread confusion concerning the difference, and implementors chalk it up to "well, photoshop does something weird but I can't figure out what". It is one of a small family of models directly related to HSL/HSV, and shares its heritage with them. To understand how hue and saturation work in HSL/HSV, it is necessary (or at least extremely convenient and common in the literature) to pass through hue/chroma first, and to understand why HSL lightness/HSV value are poor proxies for perceptual lightness, luma makes a decent object of comparison. In several prominent computer vision papers, it is compared directly to HSL/HSV. It fits in better here than in its own article, because there is at least 50-60% overlap in the required history, context, usage sections. It's possible there's some organization by which 2 separate articles would work out, but I personally don't think they'd be as useful to the "advanced" readers mentioned in the comment above (or to anyone trying to deeply understand HSL/HSV..... anyway, sorry that all these comments are kind of long; I'm not trying to blab, but just to work through some of my reasoning in constructing the article, and demonstrate that there really was careful thought behind it. Cheers. –jacobolus (t) 20:37, 10 March 2010 (UTC)

I can't really comment on broader organization at the moment, but I will comment on two other points:

1) The main problem is simply the language that is used. You use a combination of 25 cent words and poor grammar that makes it really hard to follow the meaning of the text. Take for instance the following passage:

These RGB color spaces can usually reproduce a wide variety of colors, but the relationship between the constituent amounts of red, green, and blue light and the resulting color is unintuitive, especially for inexperienced users, because human vision is psychologically organized and conceptualized in terms of other attributes such as hue, lightness, and chroma, and traditional color mixing methods – e.g. in painting – involve mixing brightly colored pigments with black or white to achieve lighter, darker, or less colorful colors.

That's a huge run-on sentence with nine words of ten letters or more each. In particular, the phrase "psychologically organized and conceptualized" is a bit hard to swallow. What is this supposed to mean to readers? Is it safe to assume that readers will recognize this as fact, or is it in need of further explanation? Further, the next sentence simply restates what is already said. I.e. sentence 1 says "xxxx is unintuitive because...", and sentence 2 says "Thus, xxxx is unintuitive." Reading this, people are going to feel they missed something in the interim.

To be accurate, the sentence is perfectly grammatical, not a run-on sentence. I'll grant you that it could be broken up and clarified, though. –jacobolus (t) 03:18, 11 March 2010 (UTC)

The next paragraph begins by speaking directly to the reader using words like "we", then ends with a conclusion that is supposed to seem incontrovertible. Are you trying to explain facts, or persuade someone? Persuasive writing is not good form in an encyclopedia. Lastly, there's a tiny bit tacked on as if in afterthought: "especially to users familiar with subtractive color mixing of paints but not with additive color mixing." Why especially? If it's so important, why didn't you explain it?

No, the goal is to demonstrate/clarify, showing through an example what is meant when sources claim that one space or another is more "intuitive" – without a concrete example, in my experience trying to explain this face-to-face, people have great difficulty figuring out quite what's going on. As soon as an example is added the lights go off and they tend to just "get it". The "we" may be too conversational in tone for an encyclopedia. It's a rhetorical device used to make the example seem more immediate. As for "especially"... someone who has experience with subtractive color mixing in general must somewhat "unlearn" the color mixture rules in order to figure out additive color mixing. I could try to track down some sources for this; at the very least there are countless books about digital imaging tools which make similar claims, but perhaps it's only anecdotal/rhetorical. It might be better to just state plainly that the model differs from subtractive color mixing without speculating about what users will find tricky. –jacobolus (t) 03:18, 11 March 2010 (UTC)

"As for "especially"... someone who has experience with subtractive color mixing in general must somewhat "unlearn" the color mixture rules in order to figure out additive color mixing." OK, then explain it in the article. SharkD  Talk  23:06, 12 March 2010 (UTC)

The third paragraph begins with, "In an attempt to accommodate these more traditional and intuitive color mixing models..." Which models? Didn't you just get done calling one of them unintuitive? Or was that another? Huh? The rest of the paragraph contains a mess of historical references that belong in their own section (along with Runge, Munsell, Ostwald, Tektronix) and not here. Be satisfied to answer the "what" and "why" of the motivation, not the "who" and "when".

No, you're misreading. The "traditional"/"intuitive" models are those based on tints/shades, or based on roughly hue–lightness–saturation/chroma, as compared to straight RGB which we just called unintuitive. As for the historical references – they refer directly to the *creators* of HSL/HSV, and their motivations in developing these models are for the most part the same reasons the models were put into software. –jacobolus (t) 03:18, 11 March 2010 (UTC)

"No, you're misreading." Yeah, that's my point. Don't write it in a way that can be misunderstood. SharkD  Talk  23:06, 12 March 2010 (UTC)

2) Terminology and abbreviations are used inconsistently throughout all the charts. The image File:Hsl-and-hsv.svg abbreviates HSL and HSV, but then uses the expanded text "Luma/Chroma/Hue" for LCH. Better to describe things that are similar in a similar fashion. Also, where is HSI? It seems to be important enough to have been factored later on into the article, so why is it absent? The phrase "vertical cross-section" also needs to be made plural as there are more than one of them, and it would also help if there were arrows pointing to each of them.

The slices seem pretty clearly tied to the text by their axis labels, which are large and clear, and by their ordering. The reason it uses "luma/chroma/hue" is that this model is nowhere called LCH. As for HSI: (a) The HSI model uses H₂ as a hue dimension, whereas this diagram is headed with H, (b) the component average is used in various related spaces other than HSI, called by all manner of names, (c) HSV V and HSL L are unambiguous, but the name "intensity" or "HSI I" or similar are much less standardized; “component average” cannot be misconstrued. “luma” is the technical name. –jacobolus (t) 03:18, 11 March 2010 (UTC)

"The reason it uses "luma/chroma/hue" is that this model is nowhere called LCH." See [1] and [2].

All those uses of LCH are referring to CIELAB with polar coordinates (L*, C*_ab, h*_ab) which is something entirely different. –jacobolus (t) 22:24, 13 March 2010 (UTC)

"The HSI model uses H₂ as a hue dimension, whereas this diagram is headed with H" So adjust the image/heading appropriately? HSI should definitely be included in the image if it's worth also mentioning in the article.

It is. Look down a bit to the diagram below. The point of this diagram is to compare possible lightness dimensions. If we had one of its four images show a different chroma dimension than the other three it would be confusing without any particular benefit. –jacobolus (t) 22:32, 13 March 2010 (UTC)

"...but the name "intensity" or "HSI I" or similar are much less standardized;" If it's too ambiguous for the image, why is it not also too ambiguous for the text?

I don't understand what you’re saying. –jacobolus (t) 22:32, 13 March 2010 (UTC)

"“component average” cannot be misconstrued." Assuming it can be understood at all. The term is not mentioned until later on in the text.

There’s an equation written directly underneath. The two words "component" and "average" are quite straightforward. And "later on in the text" in this case means "directly to the left of the image". –jacobolus (t) 22:32, 13 March 2010 (UTC)

"“luma” is the technical name." Who said it wasn't? SharkD  Talk  23:06, 12 March 2010 (UTC)

The point is, where we actually have a precise name for the dimension, which is useful independently of the rest of the model, as is the case with "luma" or "component average", we use that. Where the dimensions are absurd and artificial, we label them with HSL or HSV, because to just write "lightness" or "value" would be misleading, given that those terms are being horribly abused (i.e. they are terms which have a clear technical meaning, and it is emphatically not one of M or (M + m)/2. –jacobolus (t) 22:32, 13 March 2010 (UTC)

Next, the image File:Hsl-hsv_chroma-lightness_slices.svg uses "component average" and "luma" for HSI and LCH, but actually uses the abbreviations for HSL and HSV. The first and last should be labeled "HSI intensity" and "LCH luma" respectively. Further, maybe the individual figures should be labeled as A, B, C and D to help clarify which passage in the text refers to which figure?

See above. We could add labels, but I'm not sure they're necessary. I'd be interested to hear more people's first impressions. –jacobolus (t) 03:18, 11 March 2010 (UTC)

I think labeling things consistently is extremely important; otherwise for many readers the images will just be a bunch of noise. SharkD  Talk  23:09, 12 March 2010 (UTC)

Okay, I'm going to add a., b., c., etc. to these. Then we can properly refer to them as fig. 10a. or fig. 15c. or whatever. (So dude who's removing the "fig. «foo»" labels, can you give it some time?) Sound okay? It’ll probably take me a while to actually open the images up, make the changes, and upload altered versions, so it might not be a day or a few. Hope that’s alright. –jacobolus (t) 02:07, 13 March 2010 (UTC)

Lastly, the pictures of the three-dimensional color solids in File:Hsl-hsv_models.svg small and hard to make out. They need to be enlarged and made clearer. I think this may be the basis for some of the complaints. The old version of the article contained several images that showed the different color models side-by-side, which really helped demonstrate the differences quickly by being visual instead of textual. SharkD  Talk  22:56, 10 March 2010 (UTC)

Really? Does just enlarging the whole image a bit improve things? IMO they are currently much easier to see/understand than the previous set of diagrams were, but the cylinders at the top could potentially be enlarged a bit relative to the rest. –jacobolus (t) 03:18, 11 March 2010 (UTC)

I've expanded the 3d bits at the top by 40%. Does that help at all? If you don't like the angle & proportions of the cylinders pictured, we could swap in some of your 3d-renderings, if some of the issues w/ them described below are addressed. –jacobolus (t) 10:43, 12 March 2010 (UTC)

That's an improvement. It still looks like a hand-drawn sketch though. SharkD  Talk  23:08, 12 March 2010 (UTC)

What particularly makes it look "hand drawn"? IMO, this should show exactly the colors in the space, where they belong, without any 3d shading to (potentially misleadingly) distort them. –jacobolus (t) 01:00, 13 March 2010 (UTC)

Next, I think if H₂ is labeled using a subscript, then I also think H should be labeled H₁. SharkD  Talk  23:36, 12 March 2010 (UTC)

I prefer the current ones, but does it really make a difference? This seems like bikeshedding to me. The article is still primarily about HSL/HSV, which is why H, L, and V are just written as they are (S of course must be disambiguated). Then H₂ & C₂ are set apart as a feature which distinguishes some similar models. –jacobolus (t) 01:00, 13 March 2010 (UTC)

By the way, thanks a lot for taking a closer eye to this all, SharkD. I think for the most part your changes are a big improvement to some rather unclear sentences I had. :-) –jacobolus (t) 04:54, 11 March 2010 (UTC)

image credits in footnotes

As far as I can tell, the MOS advocates citing sources for text, facts, insights, &c. in footnotes. I think that the images in this article are as important as the facts or text, and likewise put credits for them in footnotes. Wikipedia currently has a policy against putting credits in every image caption, on the grounds (as far as I can tell... this isn't clearly elucidated anywhere) that it adds clutter. DESiegel had the idea to group the image footnotes and separate them from other footnotes. This seems reasonable enough. I realize that by the terms of these licenses, it is technically minimally sufficient (according to some lawyer somewhere, presumably) to only include a credit on the image talk page. Nowhere that I can see is including credits in footnotes precluded by any policy, and in my opinion it is more ethical, and more tasteful, not to mention more verifiable, to include footnote sources for images as well as other article sources. I have put a note to this effect on the relevant talk page, here ... maybe we could discuss it there? Cheers. –jacobolus (t) 04:46, 9 March 2010 (UTC)

See also: Wikipedia:Administrators' noticeboard/Incidents#edits_marked_with_copyrights. Perhaps I should start signing "my" articles now too. tedder (talk) 05:52, 9 March 2010 (UTC)

There's really no need for the sneering tone. It does absolutely nothing to aid the discussion. As it happens, I think it would be rather wonderful for you to spend several dozen hours making diagrams for some wikipedia article or another, or spend a few weeks crawling around in a bog taking photographs, or whatever, and then stick a credit for said diagrams or photographs in a footnote someplace. It would be of tremendous benefit to the project. Cheers! –jacobolus (t) 10:28, 9 March 2010 (UTC)

As a related note, I would really appreciate attempts at article feedback which go beyond the completely trivial content in a few footnotes. I am convinced that there are still many rough edges, ways that this explanation could be rephrased or rearranged to help readers interested in learning about HSL/HSV to understand what it's trying to say. If you (or anyone else interested in this footnote question) would be willing to give the article a read-through, and offer back your thoughts, that would also benefit the project. –jacobolus (t) 10:35, 9 March 2010 (UTC)

I think you're reading into a tone that doesn't exist. The point is that credit for photos and text are credited in the page history (and on the image page, which is a perfect spot for such a caption). Wikipedia guidelines are entirely clear on this: WP:UP says "One should never create links from a mainspace article to any userpage", and WP:CAP says "Unless relevant to the subject, do not credit the image author or copyright holder in the article. It is assumed that this is not necessary to fulfill attribution requirements of the GFDL or Creative Commons licenses as long as the appropriate credit is on the image description page. If the artist or photographer is independently notable, though, then a wikilink to the artist's biography may be appropriate." tedder (talk) 15:53, 9 March 2010 (UTC)

Okay, I'll definitely kill the user-space links. I didn't know about WP:UP. For the rest, it's easiest to keep discussion over here. --jacobolus (t) 10:04, 10 March 2010 (UTC)

Comments about restored images

• cone image: the problem with this image, and the reason I scrapped it before, and replaced it with the image currently at the top of the page, is that its shape and labels fall somewhere between misleading and inaccurate. In my opinion, it doesn't add any information to the article. The bit on the left is basically a rip off of the Painter color picker, shown later on in the article, and both parts, labeled as they are, perpetuate the common misconception that these models are geometrically conic. Wikipedia should strive to break such misconceptions.

Misleading and inaccurate how? SharkD  Talk  23:10, 12 March 2010 (UTC)

Because the labels imply that the maximum linearly saturation tails off to 0 as we approach black, but that is clearly not so from the definition. This is why I go to so much (perhaps seemingly unnecessary) effort to explain what (after Joblove & Greenberg) this article calls chroma, exactly to clear up this misconception, which is absolutely rampant, not only on the web but also in books, etc. It leads any reader who doesn't carefully read the math to have a simply wrong understanding of how the space is constructed. It's sloppy and misleading in the most generous interpretation, but yes, I’d call it inaccurate. Furthermore, the diagram below showing chroma shows a triangle, the "construction of" flow chart diagram shows the hexcone, and the user interface examples diagram shows Painter’s interface, which is just the left of this (but doesn't have the misleading labels). Sorry if I seem a bit worked up about this, but I think it actually matters to be careful about these things, so that we don’t leave people with a false understanding. –jacobolus (t) 02:02, 13 March 2010 (UTC)

Nonetheless, these representations are common. If there's something wrong with them then this needs to be handled in the article. SharkD  Talk  09:18, 14 March 2010 (UTC)

Okay, we could add a section to the article called something like “common misunderstandings” and add this image there, perhaps along with some of the terminology mixups, etc., if you want. I don’t think it’s really essential but it might be useful to someone. –jacobolus (t) 06:34, 15 March 2010 (UTC)

• two cylinder comparison image: the concerns I have with this image are as follows: (1) the image on the left has a solid white top, and therefore does not reveal the essential structure of HSL; (2) the coloring of the cylinder edges and cross sections is dramatically incorrect (compare with the image at the top of the page)... notice how the lighting on the right and left sides of each removed wedge appear to be colored differently, quite apart from any difference in hue; (3) the background grid is unnecessary and distracting; (4) The shading of the lines showing how the cylinders is partitioned are almost invisible in many places, because they don't contrast with the coloring of the cylinders themselves; (4) the words "Lightness" and "Value" are leaning over enough to appear falling over on their backs, and additionally are chopped off on the side; (5) the other words, particularly "Hue", are skewed and angled in a way which makes them difficult to read; (6) this is not especially important, but the convention in math, physics, etc. is to use the right hand rule for orienting coordinates. In other words, the positive angular direction goes counterclockwise, when the "z-axis" is pointed up out of the page towards us. I've tried to use this convention throughout (this is also the way CIELAB, CIELUV, CIECAM02, &c. will naturally be drawn, and there's some advantage to consistency with those too).

That's because there's a light shining on them from the top. I don't think readers will have any trouble telling the difference with there being environmental effects in the image. SharkD  Talk  23:10, 12 March 2010 (UTC)

Okay, well I have trouble telling the difference, and I consider myself a pretty sophisticated consumer of color diagrams. I think the lighting substantially reduces the information content of the diagram, and is gimmicky. It's a classic example of Tuftean chartjunk in action. (I'm not trying to be offensive: I really think this is a misleading diagram).–jacobolus (t) 02:02, 13 March 2010 (UTC)

Well, I disagree. SharkD  Talk  09:19, 14 March 2010 (UTC)

Do you have any reason for adding the lighting, gaps, grid background, 3-d letters, etc. other than “it looks cool”? I’m perfectly happy to have a rendered diagram, but only if we can eliminate the lighting. Otherwise, my recommendation is just to get rid of the image altogether. The point of these diagrams is to teach and impart information. Anything that detracts from that is a regression, not an improvement. –jacobolus (t) 06:34, 15 March 2010 (UTC)

"The point of these diagrams is to teach and impart information. Anything that detracts from that is a regression, not an improvement." Yes, I think we're all concerned about this. Hence all the discussion. SharkD  Talk  04:40, 16 March 2010 (UTC)

I don't mean to be obnoxious, but my impression in this discussion thread is that you're trying to dodge its subject, which was a list of concerns I have with the diagram's form and content. I'm not going to state categorically that there are no reasons for using a 3d render, but I thought I had a pretty decent list of legitimate problems, and there hasn't really been any response. –jacobolus (t) 04:53, 16 March 2010 (UTC)

OK, then. My argument in favor of shading is that what is being depicted is a three-dimensional object. Without shading and perspective you run into problems of not being able to figure out what is what, such as in the image at right. SharkD  Talk  05:06, 16 March 2010 (UTC)

“isometric flaw”

“shading 2”

“shading 3”

I'll try to rework it such that the lighting doesn't influence the coloring so much. SharkD  Talk  05:14, 16 March 2010 (UTC)

It seems the gamma in the images was set a bit high. I've uploaded newer versions and lowered their gamma. SharkD  Talk  01:11, 22 March 2010 (UTC)

I tried incorporating the images into the image you made earlier, but the SVG renderer used by Wikipedia doesn't antialias the images too well... SharkD  Talk  02:43, 22 March 2010 (UTC)

Ah, hmm. You're right that that's pretty icky. What happens if the original is rendered at a higher resolution? I wonder what their SVG renderer is, and why it uses such crappy interpolation, and whether it’s possible to reconfigure it to do a better job (it should at least use bicubic interpolation; ideally it would even use somethting better). Given that the result is cached and barely ever needs to be regenerated, optimizing for render speed over interpolation quality seems like a terrible idea. –jacobolus (t) 17:31, 22 March 2010 (UTC)

They're already rendered at a pretty high resolution (1600x1200). SharkD  Talk  11:02, 23 March 2010 (UTC)

Another possibility would be to make it an SVG file, and then render a png image at precisely the size we plan to use in the article, and then use that, putting a link to the SVG file on its image description page. –jacobolus (t) 17:31, 22 March 2010 (UTC)

Or we could use our own renderer to create a high resolution PNG version. Wikimedia's bitmap interpolation usually works pretty well. These sorts of "tricks" are usually frowned upon however. SharkD  Talk  11:02, 23 March 2010 (UTC)

To be future-looking we shoul upload at least an SVG version. Whether a high-resolution or specific-resolution png is used in-article wouldn’t make too giant a difference, though the specific-resolution version would undoubtedly look at least marginally better. A high-resolution png would let people use it at various sizes, e.g. on other articles. –jacobolus (t) 21:35, 24 March 2010 (UTC)

Okay. A reasonable enough concern. The question then becomes: for the example of diagrams showing color geometry, is that still a problem? For example, in the image at the top of the current article, are there problems figuring out what is what? Avoiding the problem in this image “shading 2” is precisely why I added a black outline. In my opinion, said outline does a fine job clarifying the geometry, while avoiding the problem of mixing up color within the shape with the shading (e.g. in your image “shading 3”, the lightness, hue, and chroma have all been altered from one side to another, making it impossible to figure out what color the shape is supposed to be in a context where that matters). –jacobolus (t) 05:28, 16 March 2010 (UTC)

Basically, the deficiencies with these images were precisely the reason that I created a new image; I'd tried suggesting some of the problems at some point months or maybe years ago, but they were never addressed and I figured it'd be easier to just remake new ones from scratch showing all the salient features of the spaces. I think both of these images are essentially redundant to the one now at the top of the article.

jacobolus (t) 02:39, 11 March 2010 (UTC)

About your Third Opinion request:

Disclaimers: Although I am a Third Opinion Wikipedian, this is not a Third Opinion in response to the request made at WP:3O, but is merely some personal observations and/or information about your request and/or your dispute.

Comments/Information: I suspect that the Third Opinion request made on this topic has gone unanswered simply because there's no regular Third Opinion Wikipedian who feels that they have the technical expertise needed to understand the matter well enough to have an opinion on the matter. (That's certainly the reason that I haven't opined.) You might be more likely to get some help if you were to make a request for comments, perhaps using the {{rfctag|sci}} tag or by asking for comments at Wikipedia_talk:WikiProject_Color. If you do one of those, please remove the listing from the Third Opinion project.

Note to other 3O Wikipedians: I have not yet "taken" this request, removed it from the active request list at the WP:3O page, or otherwise "reserved" it, so please go ahead and opine on it if you care to do so. —TRANSPORTERMAN (TALK) 13:22, 15 March 2010 (UTC)

I've left a note ([3]) at Wikipedia_talk:WikiProject_Color previously and got zero responses. I don't think the project is going to be particularly helpful in terms of responses/opinions. SharkD  Talk  04:32, 16 March 2010 (UTC)

At some point I'll probably stick a request up at WP:PR, because this article could definitely use some more eyeballs. Doing such a peer review will take a fair bit of effort from anyone who undertakes it though: for best results, they'll have to actually try to work through the math and diagrams and figure out how HSL/HSV work. Ideally, we could get a few reviewers with reasonable math & spatial reasoning skill, but little previous knowledge of color or computer graphics, and a few more with extensive color science background. I'm not sure such reviewers will be especially easy to find, but I can dream. –jacobolus (t) 05:03, 16 March 2010 (UTC)

As opposed to reviewers/Wikipedians without reasoning skills? Ouch! [Ed. Missed "spatial".] SharkD  Talk  05:10, 16 March 2010 (UTC)

I've been following this discussion/lurking for a bit. The one thing that stands out most to me is an apparent picture/diagram overload. It's easy to get lost trying to navigate between the figures and the places in the text where they are referenced. This is especially not helped by the occasional large gap caused by columns of illustrations on the right hand side. I have to agree with several of Jacobolus's concerns about the similar diagram's at the beginning. File:Hsl-hsv_models.svg seems a better representation, though the early usages of shortened symbols such as S_HSL and S_HSV can be a bit confusing. Also the cross-section showing S/L and S/V vs hues at 0 and 180 is a little confusing as well. Might be nice to but a small line stub at the center on both sides to represent the axis. As for File:HSL_HSV_cylinder_color_solid_comparison.png the lighting effect works against clarity. I agree that the conical and triangle representations are fairly common and should be explained, but as I said at the start of this comment the images for this article are very dense and overwhelm the text making it harder to follow. More of the images should be moved below and on the left side or inlined with the text as much as possible. Images that don't add clarity should be removed from the article and maybe added to the see also section. The fire breather image is a good example of bad placement, there are 5 headers/sub-headers between where the image appears and where it is talked about. PaleAqua (talk) 14:48, 15 March 2010 (UTC)

“The one thing that stands out most to me is an apparent picture/diagram overload” – is that just a first impression, or do you think it's still a problem after trying to read the article? The reason I added so many diagrams & pictures is because a subject like color and color geometry is easiest for me to really understand visually, and I expect the same is true for others; Unlike with most articles, I think of the images here as a core part of the page’s exposition. As for the fire breather image: it's a direct illustration of the section it is in. You're right that it's also referred to later on... I'm not sure there's a good fix for that. –jacobolus (t) 05:09, 16 March 2010 (UTC)

The problem is that the article doesn't properly lead into the technical stuff. It's all a bit too much to swallow all at once. SharkD  Talk  05:16, 16 March 2010 (UTC)

Also, it's too easy to get confused if you are constantly switching terminology as well as the starting points in the images. For instance, in File:RGB_farbwuerfel.jpg the reference color is teal. In File:Unintuitive-rgb.png it changes to orange. In File:HSL-HSV_hue_and_chroma.svg it changes to somewhere between blue and cyan. In File:Hsl-hsv_chroma-lightness_slices.svg and the later images it is now a more pure blue. Constantly switching things around is equivalent to playing a game of three card monty with the reader. (Which pretty much sums up my opinion of the current state of the article.) SharkD  Talk  05:22, 16 March 2010 (UTC)

Hmm. Well the other side of that is that using varied examples helps with fluency: different parts of the space have different problems/different properties. Ideally we could work in a purple example too :-). The whole point of the “disadvantages” section is that these geometries are not uniform with respect to human vision. I'd be glad to have a different image of an RGB cube; I just wanted to re-use existing parts to the extent possible. The teal example there (or any specific color example) is unnecessary, for our purposes. Three card monty? We're not hiding anything. What do you mean “constantly switching terminology”? –jacobolus (t) 05:39, 16 March 2010 (UTC)

"Hmm. Well the other side of that is that using varied examples helps with fluency..." Doubtful. SharkD  Talk  01:14, 22 March 2010 (UTC)

"first impression, or do you think it's still a problem after trying to read the article?" It was both a first impression, and a difficulty in reading the article. Granted I'm understood the topic fairly well beforehand, but I still found myself having to jump around a bit trying to match up pictures with the text. I've also tried to read the article pretending that I knew absolutely nothing about the topic, and that's where I see the biggest issues. As SharkD said there is also some problems leading into topics. For example, R, G and B are mentioned as symbols before any direct mention of red, green or blue. There is a mixture of explaining what some symbol stands for and some are just given trusting that the reading will figure out the abbreviations. I actually find it slightly humorous that the introduction specially declares B is brightness but doesn't do that for any of the other cases, and then goes on to state that neutral colors are R = G = B later without stopping to explain what is meant by B has just changed. Some another examples, H = 60° · H' just below the piecewise definition then H = 60° × ... in the picture to the right. Or some pictures using S_HSL compared to HSL S etc. The sequence of figures 6, 7 and 8 feels awkward. 6 shows H, while 8 shows H₂ while stuck in the middle is figure 7 which represents the projection. What is p' in figure 6? Why is it labeled? Why is p labeled in 6 and not in figure 8? Yes someone reading the article can figure it out but as it is it takes more work that it should. Oh and figure numbering is inconsistent. I wonder if reference grouping or something could somehow be used to automate the figure numbering as well as match the numbers to the text. PaleAqua (talk) 09:51, 16 March 2010 (UTC)

Okay. I just added a “basic idea” section at the top, that avoids any discussion of RGB values and describes HSL/HSV on their own terms. Does that help? (Please feel free to rework any parts of it that seem confusing or wordy!) I scrapped the R = G = B bit from the lead section since it can be and is described later on. I’ll try to go through and make sure every mathematical symbol is defined before its use, but if you notice any inconsistencies, definitely point them out. Is the H′ thing a problem (or is the problem the particular multiplication sign?)? I thought the math worked out a bit clearer (it's more concise, anyway) with the intermediate H′ in range [0, 6), afterward multiplied by 60°. A footnote explains that H′ can be thought of as the hexagonal-angle analog of radians, 6 instead of 2π because the circumference of a hexagon of radius 1 is 6. I think that's too much information for the main body text, and would just add clutter, but maybe not? As for the sequence of images 6–8, do you have any suggestions for clearing them up? The problem is maybe that the way I think of it is that H as the primary attribute we’re interested in, and H₂ is an alternate very similar attribute used in some niches. Thus I try to give H the spotlight, but want to at least mention H₂ because it’s sometimes relevant. Re: declaring B for brightness, is that really unclear? The others say (HSL stands for ...., HSV for ...); we could say “(HSB for hue, saturation, brightness)” if that would help things. I was just trying to avoid unnecessary redundancy. Anyway, keep the review coming! –jacobolus (t) 18:32, 16 March 2010 (UTC)

Basic idea section looks good and definitely helps. I only pointed out the B as in relation to the R = G = B appearance. By itself it's fine. For the H = stuff, I was actually pointing out the two different multiplication symbols being used, raised dot (•) vs ×. For the layout of 6, 7 and 8, I mostly noticed the problem when the browser is on the wide side fits, but it fits much better when I'm using my normal browser with a narrower window. (I do most of my watch list scanning in a browser built into the RSS reader I use which tends to be fairly wide.) Perhaps if one of figures 6 or 7 was on the left it might flow better when wide, though that might make it worse in the case the browser window is narrow. PaleAqua (talk) 22:04, 16 March 2010 (UTC)

Yes, this is a huge problem with wikipedia in general. Personally, I think they should put a max-width: 45em (or even 40em) on body paragraphs. It would do a world of good for layouts generally, making the site actually readable for people with browsers wider than about 950 pixels. Oh well. I’m not sure there’s any layout here that will suit any width browser window. –jacobolus (t) 22:23, 16 March 2010 (UTC)

I agree that max-width would be nice. PaleAqua (talk) 23:33, 16 March 2010 (UTC)

Oh, also, I'm going to try to make a diagram showing H/L/C and H/V/C, with the label Chroma on the radial axis, to replace the diagram that’s currently in that section; I think the color picker triangle is adequately covered by the screenshots later on. –jacobolus (t) 18:36, 16 March 2010 (UTC)

Re: P and P′: Chroma is defined as the ratio OP / OP′, which is why we label O, P, and P′. –jacobolus (t) 18:42, 16 March 2010 (UTC)

Yes, though the article doesn't directly state what O, P and P' are, though I guess it's not really necessary. PaleAqua (talk) 22:04, 16 March 2010 (UTC)

It says “The chroma is the proportion of the distance from the origin to the edge of the hexagon. In the lower part of the diagram to the right, this is the ratio of lengths OP/OP′, or alternately the ratio of the radii of the two hexagons.” – is that not clear enough? We can add another sentence if it would help. –jacobolus (t) 22:11, 16 March 2010 (UTC)

I saw that. It's hard trying to read as if I didn't understand the subject, not sure if I'm trying to much to read at a basic level. It's obvious to me that O = Origin, P is the projection of the color from RGB space onto the hexagon and P' is the projection from the origin through P onto the outer hexagon and that the notation OP and OP' refer to the lines between those points... I'm not really sure how much basic knowledge is okay to assume and how much should be explained. Guess leave it alone for now. PaleAqua (talk) 23:33, 16 March 2010 (UTC)

Re: figure numbering: I’d love it if there was a way to assign names, and have the figures auto-number based on order, and then fill in the proper numbers for named references. This is a feature of CSS3 not yet AFAIK implemented in browsers (though I'm not sure it'd be usable for the element-specific CSS allowed for general wikipedia pages). There may be some way to get mediawiki to do it – after all, they do footnote references properly – but I haven't heard of it. If you can figure it out, please please tell (or even just swap them in!). –jacobolus (t) 18:46, 16 March 2010 (UTC)

I've been wondering if the footnote number system could somehow be "abused" to get something workable, I'll see if I can figure something out. PaleAqua (talk) 22:04, 16 March 2010 (UTC)

a few observations

I've been recently reading through most of the color articles when I came across this page. As a disclaimer, I have no previous experience with color models, besides what I've read in the past couple days. I feel that after reading this article, I have a pretty good grasp of HSL and HSV, or at least how they compare to other systems, so congrats on that part. Some general comments:

The intro is rather long. This might not be helped... as it is a pretty complex subject if you don't have any background. However, lines such as "Unfortunately, while typically consistent, these definitions are not standardized, and any of these abbreviations might be used for any of these three or several other related cylindrical models." could be written more concisely (bit of a nit-pick, but the shorter it is, the easier it is to comprehend on the first reading).

I'd also stay away from a redlink in the intro, though it appears you might be waiting to write this article at a later date. I think there's a comma missing in the last sentance (I'm horrible with spelling/grammar myself so I'm not positive).

The image at the right is very nice (and I can tell you've revised it several times) but the caption fails to adequately explain the figures. After reading the entire article, I can tell what you mean by "keeping the variable fixed" but someone who's just looking at the pictures (upon first finding this page) would be better able to understand if it were written in terms geometric surfaces and cross sections. For example you could say that the second set of images represents the rectangular face of the cylinder (unwrapped), the third set of images represents a horizontal cross section, and the last set of images represents a verticle cross section. I realize this is self-evident if you know what the labels of each image mean, but in the intro of the article, that could be assuming too much. If you do take to time to edit this image again, I would also darken the labels on each figure as they blend into the grey background. You might also exaggerate the white at the top of (b). That's a lot of critique for one image but it's so informative that it might need some more explanation.

I don't really get what the Tektronics image is for. What does this device do? Is it included because it was the "first" biconical representation of color? Was it the direct inspiration for HSL/HSV? I think those questions could be answered pretty easily in sentance or two.

You might want to expain the angle of the rotated cube and how it applies to the different colors' inherent lightness/luma. Might want to explain how Rec. 709 came up with it's luma values or that could be too much for the article. Either way, it seems arbitrary IMO.

I agree with the previous editor's comment about the firebreather picture. While it might be appropriate, I did end up skipping it until the later section where it referenced when I first read the article.

I think it may have too many references to sound-alike variables, but again, don't know how you can resolve this. L, L*; H, H₂ are the two that threw me. But that could be because I didn't really attempt to follow the math too closely.

" For instance, rotating the hue of a pure dark blue toward green will also reduce its perceived chroma, and increase its perceived lightness, but the same hue rotation will have the opposite impact on lightness and chroma of a lighter bluish-green –   to  . In the example below (fig. 20), the image on the left (a) is the original photograph of a green turtle. In the middle image (b), we have rotated the hue (H) of each color by −30°, while keeping saturation (SHSV or SHSL) and value (V) or lightness (L) constant. In the image on the right (c), we make the same rotation to the HSL/HSV hue of each color, but then we force the CIELAB lightness (L*) to remain constant. Notice how the hue-shifted middle version without such a correction dramatically changes the perceived lightness relationships between colors in the image. In particular, the turtle’s shell is much darker and has less contrast, and the background water is much lighter."

I understand the physical description, e.g. one is lighter than the other, but it is confusing to switch back and forth between models. For example a hue shift would imply HSL/HSV because of teh evenly spaced colors. However chroma isn't a component of either, so it isn't implicit (at least to me) why it's value changes (or the lightness, which I would specify which lightness you mean) or it could all be because of the "percieved" qualifier before each term. When you switch over to CIELAB lightness is that equivalent to perceived lightness? How would you go about making a 30deg hue shift in a CIELAB model (cartesian coordinates?) I guess I'm not sure how this section explains a phenomenon vs. just presenting it.

I didn't read the other models section or look over the math.

I know this is a mouthful of critique but it looked to me like you two were going back and forth with little 3rd party imput (or reluctant imput). I hope this helps from the perspective of a layperson. I'm impressed you've been able to keep it civil :) it's been somewhat entertaining reading the back-and-forth comments. Lime in the Coconut 20:15, 19 March 2010 (UTC)

Thanks for taking the time to read/comment! I’ll try to break this up point-by-point so it’s easier to follow. Feel free to just stick responses after one of the points if you want.

“The intro is rather long.” — Yes, I agree. I’m not sure exactly what the best amount of information (or which specific bits) is most important for the intro.

“[the first image caption] would be better able to understand if it were written in terms geometric surfaces and cross sections.” — Okay. That’s probably a good idea. I was trying to keep the caption brief, and let the image speak mostly for itself, but it’s easy for me to forget that lots of people just skip any equations they see altogether. :-)

“I would also darken the labels on each figure as they blend into the grey background.” — That was the whole idea. :-) I tried with dark labels, and the image already has enough stuff on it (especially enough fairly dark stuff) that the clutter was soft of overwhelming. The labels aren’t especially important as content; I think as long as they’re dark enough to be legible, it’ll work okay.

“You might also exaggerate the white at the top of (b).” — As in, add a darker grey line behind it? That’s maybe reasonable. It's the approach taken in figs. 11–14. I’ll try adding an outline just to the top of 1b but not to its sides or c, d, f–h, and hopefully no one will think that there's actually a dark grey top to that cross-section. :-)

“I don't really get what the Tektronics image is for.” — Yeah, Tektronix made expensive “graphics terminals”, and came out with one whose color designation was based on HSL in 1979, the first real implementation of HSL (albeit with hue rotated so blue was 0) that I know of. Maybe including the image is just fetishizing the historical, but I love old drawings and diagrams like this, and this particular one is in the public domain as part of a patent.

“Might want to explain how Rec. 709 came up with it's luma values” – I think that’s out of scope for this article; if you read the page about luma, or the references, for instance Poynton’s “YUV and luminance considered harmful: A plea for precise terminology in video”, it’s explained in detail.

“too many [...] sound-alike variables [...]. L, L*; H, H₂ are the two that threw me.” — Yeah, it’s hard to get people up to speed with giant hunks of color science in the scope of an article like this. :-) The idea is that CIELAB L* is a pretty good model of the “lightness” that people actually see. HSL L, by contrast, is not. But to understand what L* is requires learning a bit about how CIELAB works (unfortunately the Wikipedia article about it is currently awful). There's a quick description of it in the section about “other models” at the bottom, but that part really needs a picture and another sentence or two about why CIELAB is “pretty good”. H and H₂ are nearly identical (at most they differ by ~1.1°), but derived using two different formulas. Basically, H is what you get if you pretend that the hexagonal “chromatic plane” projection of the RGB cube is really a circle, not a hexagon (i.e. warp the hexagon into a circle shape), and H₂ is what you get if you leave it as a hexagon and just measure with respect to polar coordinates in that plane.

“For example a hue shift would imply HSL/HSV because of the evenly spaced colors. However chroma isn't a component of either, so it isn't implicit (at least to me) why its value changes (or the lightness, which I would specify which lightness you mean) or it could all be because of the "percieved" qualifier before each term. When you switch over to CIELAB lightness is that equivalent to perceived lightness?” — As you surmised, it’s all because of the “perceived” before the term. The idea is that if you just look at the picture, parts *look* lighter and more/less colorful. CIELAB lightness is a much better model of perceived lightness (this is pretty clear in the fire breather image, for instance). I’ll try to look back sometime and think about ways of making this clearer. If you have any ideas, I’m all ears. :-)

Here’s an analogy: if you have a hot water knob and a cold water knob on your shower, and you turn one of them, you’ll change both the temperature and pressure of the water coming out of the shower head. Usually, you just want “more water” or “cooler water” or similar, but to accurately accomplish that change, you need to turn *both* knobs. Well, trying to change perceived lightness (like our “more water” in the shower case) you have to fiddle with both V and S or both V and L. But it’s worse than the shower example, because the adjustments you make to do what you want to do to blue might do something completely different to green, for instance.

Thanks again! –jacobolus (t) 20:26, 20 March 2010 (UTC)

You should also use a Latin/Roman font for the equations, just like the standard LaTeX output. SharkD  Talk  02:51, 22 March 2010 (UTC)

It’s not necessary (by the manual of style; to whit, “Either form is acceptable, but do not change one form to the other in other people's writing. They are likely to get annoyed since this seems to be a highly emotional issue.”) and I dislike how the html versions of math tags render. Just using italicized text is fine. –jacobolus (t) 16:37, 22 March 2010 (UTC)

>Sigh< No, you don't have to do it... I still think it would be better. SharkD  Talk  18:03, 26 March 2010 (UTC)

Okay. Well, I think it would be much much worse, and I cringe every time I see an article that uses that version. On general principle, I’d like to use a font with serifs, too, but the specific html math rendering done by mediawiki is just bad looking (wrong sizes, wrong alignment, bad spacing, ugly glyphs for greek, poor superscripts/subscripts, etc. etc.). –jacobolus (t) 18:11, 26 March 2010 (UTC)

pov-ray code for the 3D HSL image at the top

SharkD: If you want to play with a shape approximately like the one in the top image, here's some pretty close POV-ray code. HSV would be pretty similar. –jacobolus (t) 23:11, 20 March 2010 (UTC)

Image labels

Before I get into picking at some details just wanted to say I really like the images that you all have made. And the new bicone / cone figures in the basic idea section do look really nice. PaleAqua (talk) 09:11, 22 March 2010 (UTC)

In File:HSL color solid cylinder alpha lowgamma.png and File:HSV color solid cylinder alpha lowgamma.png the "Hue" label vanish against the dark colors over which they appear. Might be better to either spin the arc backwards a bit so that it appears to come from behind the cylinders (which would be the case if it started at red / 0 ) and that the text is over the grey to the side, or switch the color of the text to white for contrast. The "Saturation" label also has a little bit of a contrast problem but not as badly. PaleAqua (talk) 09:11, 22 March 2010 (UTC)

Yes, I noticed exactly this myself. I am trying different ways of fixing it. SharkD  Talk  00:51, 23 March 2010 (UTC)

In File:HSL color solid dblcone chroma gray.png and File:HSV color solid cone chroma gray.png the labels have better contrast the but the lines and arcs seem a bit loose. For the HSL hue arc it almost seems like it is pointing downwards a bit towards the base, rotating it back a little bit towards orange and away from green so the that arrow head doesn't touch the base might help here. I realize that the labels in the HSV image are in the exact same location in all the images, but they appear to be misaligned against the color solid. HSV hue seems to be floating a bit off and have a center that doesn't match the center of the cone. Likewise HSV chrome seems to not visually line up with the edge of the cut out. Looks as if left end of the arrow is further from the edge than the right end. Finally the value and lightness lines seem to be angled outwards slightly. While this actually works for the cylinder versions since the edges of the cylinders reveal the projection with the bicone and cone the only vertical line is the axis so even if the 2D projection is actually right, it seems off. One possibility I suppose would be to move the value and lightness labels to be against the axis with the label on the yellow green side, though that would introduce contrast problems. PaleAqua (talk) 09:11, 22 March 2010 (UTC)

The labels right now are in a plane parallel to the equator. I have some ideas already on how to tweak them. SharkD  Talk  00:53, 23 March 2010 (UTC)

Some comments/questions on these new images:

• I really like the RGB cube. Much better than the previous one. (though again, the lighting isn’t my favorite)
• SharkD, how attached are you to the flat in-perspective label text? The arrows work fine for me in that style, but I’d prefer to just stick the text over the top, in a straight orientation and un-skewed (I find the "hue" labels, or, for example, "green", really hard to read. When you're done getting the image part the way you want, do you mind if I just stick the text over the top in Myriad font (the rest of the diagrams use that).

I very much prefer it that way. SharkD  Talk  01:00, 23 March 2010 (UTC)

• I'd be happy to stick whatever final 3d cylinders we can agree on straight into that svg image at the top. I think that'd work better than the alternative, which is to make that image just have the 2-d slices, and leave the 3-d bits in a separate image stuck above it.

I agree. However, the SVG renderer used by Wikipedia isn't up to the task. The antialiasing (edge blurring) is pretty non-existent as far as I can see. SharkD  Talk  01:00, 23 March 2010 (UTC)

• I’m still not super happy with the lighting in either the cones or the cylinders: they don't appear to actually go to white at the top in the cross-section surfaces to the left. It makes it look like the left side is substantially darker than the right. Also, the grids going through them, while spiffy, are made variably visible by the lighting (especially the top of the cylinder has some particular bits of the grid completely visible and some other bits that completely melt into the background.

I'm going to try rotating the models a bit. More of the outer surfaces will be showing. SharkD  Talk  01:00, 23 March 2010 (UTC)

• In HSL, do you think that the current cross-sections actually show what’s going on? The hiding of the circle passing through the “pure colors” (i.e. red/yellow/green/etc.) by the rest of the solid obscures an important part of the structure. I'm not completely satisfied with my solutions to this, either.... PaleAqua and others, any ideas?

I don't think this will cause people to trip. But maybe I'm wrong. SharkD  Talk  01:00, 23 March 2010 (UTC)

Especially in the case of the HSL cylinder, the giant pure white top takes a lot of space to represent very little information, while the overall structure is obscured. People can figure it out (esp. by comparing to the separate cross sections, etc., but it’d be a better summary if they didn’t have to I think. The same is somewhat/mostly true for the hue/chroma/HSL lightness bicone. –jacobolus (t) 21:39, 24 March 2010 (UTC)

–jacobolus (t) 17:23, 22 March 2010 (UTC)

For the cross-sections, how about showing a small guide image illustrating where they are from. For example a cylinder with the top half flipped open showing the circle of colors. See [4] for an uncolored example. Likewise I can image the cylinder front side open to reveal the saturation/lightness or saturation/value cross sections. For the outer edge, maybe something that looks like peeling the edge of a cylinder? PaleAqua (talk) 00:30, 23 March 2010 (UTC)

I'll try to come up with something. SharkD  Talk  02:33, 23 March 2010 (UTC)

I was unable to recreate the cross-sectional images, as I don't know the formulas involved. SharkD  Talk  01:17, 28 March 2010 (UTC)

Ideally we could make some animations (I can’t think of how to easily to make a 2d static image showing this – others should feel free to try though!), but unfortunately using GIF as a file format doesn’t work for diagrams needing lots of colors. –jacobolus (t) 04:38, 28 March 2010 (UTC)

Updated versions of images

OK, I've uploaded new versions of the files. Refresh your browser cache and see if they are any better. SharkD  Talk  05:12, 23 March 2010 (UTC)

Looks much better. PaleAqua (talk) 10:02, 23 March 2010 (UTC)

• Slanty labels are still annoying, particularly "Hue" and "Green" which need some change or another, and "Lightness"/"Value" which appear to be falling over backwards, but to some extent the others too. All the labels are slightly too close to their arrows, and the hue arrows are too close to the solids. If you don’t have a copy of the Myriad font, I’d be happy to email you one, or run your povray code here, if you think it'd be helpful (for instance, real fractions look better than 1/5, etc.). The font probably doesn’t make or break anything though. Or if you make a version without text, I could stick some nice upright/unskewed/readable labels over the top.
• I’ll lay off harping on the lighting, but I don’t think it’s helpful. Maybe we could compare 3-4 variations in lighting for one of these, and try to come to some consensus about the best balance between "nifty looking" and "shares accurate information".
• I prefer an orthographic projection but for this it doesn’t make a tremendous difference. The projection ends up being extremely frustrating in Apple’s ColorSync Utility (for viewing the geometry of ICC profiles), because top or side views end up inaccurate but that’s not made clear to users, and ends up misleading them.
• The seam that goes straight up from the center (in the tops of cylinders/cone) is a visual distraction. Can we rotate these just enough so that it’s slightly angled?
• I don’t think putting the 3 images side-by-side is so great. The text ends up stretched into tall narrow columns and the tint/shade diagram is completely unreadable at that size.

–jacobolus (t) 21:29, 24 March 2010 (UTC)

Wow! At $229 that would be a very generous gift! SharkD  Talk  04:41, 26 March 2010 (UTC)

I retract my offer. You’re right, Adobe might someday decide to swoop down and sue everyone who has ever emailed one of their fonts to a friend (especially one that they have distributed with every version of their ubiquitous software for the last decade). I retract my offer to illegally share a $35 piece of intellectual property. As an upstanding member of society, I of course would never do such a thing. I’m going to assume your request for a similar breach of copyright a while ago was also just a bad joke, right? right? –jacobolus (t) 17:51, 26 March 2010 (UTC)

I spent about an hour drawing this picture trying to show what I think are problems with your figure, and have tried to be concrete and specific. If the only response you have, in a week, is a sarcastic joke, how are we supposed to come to any kind of consensus about it? –jacobolus (t) 19:06, 29 March 2010 (UTC)

A specific idea for one potential fix of the “hue” label: what happens if you make the arrow wrap around a vertical cylindrical shell, rather than lying on a circle in a plane? Then the label would be vertical rather than horizontal. –jacobolus (t) 19:52, 26 March 2010 (UTC)

Meta-conversation about editing and communication style

For future notice, please do not edit other people's comments on Talk pages, nor your own significantly after the fact. SharkD  Talk  17:41, 26 March 2010 (UTC)

Listen bub, you’re being extremely passive aggressive (at least as I read it, maybe I’m misinterpreting – here on this talk page, in edit summaries, in the edits themselves). I don’t know if you’re trying to intentionally pick a fight – I'm trying my best to be patient with it – but please stop. It’s really really really annoying. Instead of having a talk-page fight about absolutely nothing useful, can we actually make substantive progress? You still haven’t made substantive answers to any of my criticisms here. Meta-meta-meta discussion is not a good use of time for either of us, and is utterly useless to everyone else who might read this page. –jacobolus (t) 17:51, 26 March 2010 (UTC)

And you have been very difficult and disagreeable throughout this entire transaction. You are simply demanding changes to every trivial matter, and unwilling to budge on any of your own ideas. This is not nourishing toward collaboration. The insults are very troubling, and I implore that you now stop. SharkD  Talk  18:27, 26 March 2010 (UTC)

Okay. As I see it, most of the “budges” are irrelevant stylistic changes where either form is widely accepted by the MOS (for instance, attempts to change spaced en dashes to em dashes, or to get rid of serial commas, or to add parentheses around the inline color samples, or to use a different font for math variables in running text, moving an image from the place where it’s most relevant to a place where it’s slightly less relevant, insistence about adding letter labels to the part of figures, or minor grammar rearrangements that result in sentences as or more awkward than the original). My skepticism of such changes is not an attempt to be “difficult”, and if you noticed I’ve been quite happy to “budge” in several cases – indeed, the majority of cases, and even willing to put in a few hours implementing those “budges” in images – where I think the change is either an improvement, a wash, or so trivially worse that putting it back or arguing about it doesn’t make much difference.

I don’t know what these “trivial” matters are that I’m being so insistent about, but I think the content and form of images is pretty important (readable labels are, frankly, an essential attribute of diagrams like these, so yes, I’ll demand a change to labels as long as they are unreadable).

If you honestly aren’t trying to be passive-aggressive (perhaps the wrong word: maybe snarky, nit-picky, argumentative, intentionally obnoxious would be better labels for the way it comes across to me), I’d appreciate if you attempt to discuss your concerns on this page frankly and forthrightly. My goal is not at all to “insult” you, and I apologize if you interpreted it that way. I’m just trying to bring to your attention the way it appears to this particular observer. Maybe I’m wrong, or paranoid, and you’re not trying to “push my buttons”... such are the problems with text as a medium of conversation – nuance and tone of voice are hard to convey. Can you at least see how I get that impression? We’re having a serious miscommunication here, and I really hope we can get on the same page, because it (i.e. miscommunication) is destructive and pointless and a huge waste of time. If you want to take this discussion off the talk page, where it might be hard to get over, feel free to email me directly. I’d email you, but I can’t find your address anywhere. –jacobolus (t) 18:43, 26 March 2010 (UTC)

Anyone can feel free to ignore this section; no need to spill more ink about it. SharkD: when you get the chance, I would appreciate responses about the images, in the previous section. Does that demonstrative image show well enough the things I was talking about? –jacobolus (t) 05:41, 27 March 2010 (UTC)

Just as you have accused me of bikeshedding[5] and dodging the subject[6], there are a number of issues you have not been willing to address yet:

• It would be better to show three-dimensional models next to the two-dimensional cross-sections, as commented upon by myself, Lime in the Coconut and PaleAqua.[7][8][9]
• All the charts and diagrams are using different RGB colors as a starting/reference point when they should instead use the same one.[10][11]
• The labeling of parameters and results in equations is inconsistent throughout the article.[12]

The best you've been able to come up with so far are snarky comments regarding chartjunk[13] and lack of sophistication among other readers.[14] SharkD  Talk  01:57, 28 March 2010 (UTC)

“bikeshedding” — Do you disagree that it was bikeshedding w/r/t the points under discussion there? I never saw any response.

“would be better to show three-dimensional models” — Okay, you claim to not know how to make such models in software. I could probably figure out a way, but I think it would take several hours of work at least for marginal benefit, and don’t want to dump time into that when many other things would be more productive. Also, neither Lime’s nor PaleAqua’s comments had anything like what was in yours. I’m not sure how you’re grouping them together. I didn’t answer PaleAqua because I have no idea how to accomplish what he’s she’s looking for. I tried to answer Lime at great length (I hope he comes back and responds again).

“All the charts and diagrams are using different RGB colors” — I answered this several times, without any further responses. (1) The hues/specific colors chosen were chosen for different criteria, which suggest different good demonstrations. (2) Having a diversity of examples is helpful because it gives readers more familiarity with the whole of the structure. (3) The actual benefit of such a change would be marginal at best and would take several hours of painstaking effort to pull off. I don’t see any kind of consensus about it, other than your repetition.

“labeling of parameters and results in equations is inconsistent” — (A) Several changes were made to the article itself in response to that, and I responded at length and specifically to the remaining points. What still bothers you?

“best you've been able to come up with so far are snarky comments regarding chartjunk and lack of sophistication” — (1) Calling things chartjunk is not a snarky comment. It is an attempt to assign a more understandable label for a whole set of problems (that is, purely decorative aspects of an infographic or chart which contain no content). (2) I only said that if I have trouble with it, as someone who has spent a great deal of time looking at and trying to interpret such diagrams, then other users with less experience or familiarity would, it is my guess, also have difficulty, perhaps more difficulty. That is not in any way to disparage such readers or suggest that they are “unsophisticated”. I doubt anyone who followed the discussion here or to the changes to the article could find this summary – “best you’ve been able to come up with” etc. etc. – credible. Your caricature is offensive. –jacobolus (t) 02:12, 28 March 2010 (UTC)

Here's my proposal: for each concern you still have about the actual content of the article, make a new section on this talk page, and we can try to work through it. I’ve really done my best to answer everything that’s come up, but it’s hard to work all the way through something and make sure nothing slips away when each comment discusses 10 separate things. –jacobolus (t) 02:42, 28 March 2010 (UTC)

This is a good approach. — e. ripley\^talk 03:04, 28 March 2010 (UTC)

Slightly off topic, I'm a girl, so should be she above for me. As far as my "requests" my main interest/what I'm looking for is seeing that the article is improved, especially in terms of it's approachability to someone not familiar with the topic and trying to make an attempt at offering another opinion which seems to be needed. I know that creating the images is a lot of tedious work, so if my comments are not seen as helpful or not worth the effort, don't worry I understand. I'm going to skip commenting on the conflict itself other than to say that I believe that both Jacobolus and SharkD have been doing great work here, and hope they can take a breather and get past this. PaleAqua (talk) 04:23, 28 March 2010 (UTC)

All the comments are definitely appreciated – I just couldn’t imagine quite how to solve the diagram question you raised. –jacobolus (t) 04:43, 28 March 2010 (UTC)

One last thing: calling in the “breach of wikiquette” police doesn’t help anything. If you feel specifically insulted, point it out and I’ll try to explain. Nowhere has my goal been to insult you. I’m still convinced that we can have a discussion like reasonable human beings, but it has been my impression that you’re trying to just pick a fight rather than solve content disputes. If that is not the case, making accusations accomplishes just the opposite. –jacobolus (t) 02:47, 28 March 2010 (UTC)

Without making any judgments specifically about either of you or your editing, my immediate observation is that you both need to focus on article disagreements instead of picking at each other. Focus on content disputes, not personalities. If your questions or suggestions are going unanswered, it may help to consolidate your dispute a bit -- pick one piece to focus on, then move on to the next. Also you should browse through dispute resolution options to see if any may fit; perhaps seeking a third opinion or opening a request for comments. This talk page is rather dense and the subject matter opaque to me personally but I'll watchlist the page and weigh in if I feel I can contribute something. In the meantime, take a walk, deep breath or whatever you need to do, then come back and focus on making an encyclopedia. — e. ripley\^talk 02:49, 28 March 2010 (UTC)

Please see Wikipedia:Mediation Cabal/Cases/2010-03-14/HSL and HSV. Would you be willing to participate in mediation? I don't see this article proceeding further without it. SharkD  Talk  10:38, 29 March 2010 (UTC)

Got it, that's a good step. I am flattered by your request but I don't think that I have the time at the moment to take on an active mediation role, and anyway I see that someone else has agreed to take on the article, however I would be more than happy to watch as things progress and weigh in as I can as a third (or I guess fourth) party. — e. ripley\^talk 11:28, 29 March 2010 (UTC)

SharkD: I can’t adequately respond to your vision of what this article should be until you clearly articulate what that is. Just stating that you think inexperienced readers will have trouble with it is a good start, but if you can’t put an image into my head/others’ heads of just what it is you mean and expect, it is, indeed, impossible to “proceed further”. I really really really am doing my level best to answer every point that comes up, and explain my motivations for writing things the way I did, in as much detail as I can. The reason I’m frustrated with you is that you keep implying that there weren’t conscious choices made, that I was arbitrary or incoherent, but without either getting down to specifics, or actually engaging when I try to explain what the choices were, and what my position is. Basically, the pattern has been, from my perspective: (1) you rush in with a quick vague snipe of a criticism, (2) I try to explain myself in (almost excessive) detail, (3) you disengage and move on to something else. I don’t think a mediator will be able to help with this, unless s/he can convince you to clearly and fully explain yourself and then stick around for the hard work of actually hammering out a solution. Finally, I don’t think we’re going to make any progress until you try to understand where I’m coming from, and either agree with me about the nature of this frustration, or else try to help me see why it is unfounded, so that we can get over it. From my perspective, you both ignoring content discussion, and pointedly denying that I have any basis for my current understanding. The pattern really pushes my patience. –jacobolus (t) 18:57, 29 March 2010 (UTC)

(Back over to the left) Let's do this -- let's take a point and talk it over. Shark, would you start? Pick a sentence or point that you want to change, that you've had difficulty seeing eye to eye about, and let's discuss it. — e. ripley\^talk 21:59, 29 March 2010 (UTC)

Example colors used in diagrams

OK, here goes. The chart images use different colors before conversion. For instance, File:RGB_farbwuerfel.jpg uses teal, File:Unintuitive-rgb.png uses orange, File:HSL-HSV_hue_and_chroma.svg uses a color somewhere between blue and cyan, File:Hsl-hsv_chroma-lightness_slices.svg uses a more pure blue. I think that all charts should use the same color before being converted (to HSL, HSV etc.) SharkD  Talk  07:20, 30 March 2010 (UTC)

OK. Why is it that you think that should be done? — e. ripley\^talk 12:49, 30 March 2010 (UTC)

I believe unexperienced readers will have trouble interpreting the material if there are too many differing variables introduced in the article. SharkD  Talk  04:28, 31 March 2010 (UTC)

I failed to notice earlier that the very top image File:Hsl-hsv models.svg uses yet two more colors: orange (I'm assuming not the same orange as the later image) for the 3D "slice", and cyan/red for the 2D cross-section. SharkD  Talk  02:53, 2 April 2010 (UTC)

RGB_farbwuerfel.jpg is not currently used anywhere. –jacobolus (t) 23:42, 30 March 2010 (UTC)

Response:

I’ve tried to explain this at least a couple times on this talk page, with pretty much no evidence that my explanations were understood, or even read. I’ll try again, though, and try to put the whole argument in one piece, as clear as I can make it. This time, I’d appreciate some acknowledgment that I’m not just talking to myself. These diagrams, in total, represent somewhere between 50 and 100 hours of effort. None of the choices in their construction was arbitrary, and great care was taken to make them as clear as possible. (Which is not to suggest they couldn’t be improved, but only that one-sentence demands for changes that I disagree with, unbacked by evidence, persuasive reasoning, or even a nominal effort to understand the trade-offs involved, don’t really count for much, even when repeated like a mantra.)

The exemplary colors in the various diagrams in the article have different goals which suggest different criteria for their selection. The expectation is that the typical reader can see in color, and can read the text, and will be trying to use a diagram to aid comprehension of the text, numbers, and mathematical formulae. To be useful, each diagram demands more than a glance, and must be taken as its own object; some diagrams might take a few minutes to fully appreciate. These diagrams are spaced out over several pages worth of article. In summary, these goals suggest that the clearest example color be chosen for each chart, for its own sake, because a reader trying to comprehend the chart will be looking at the relations demonstrated within it, rather than at another unrelated diagram several sections away.

So, the specific goals:

The diagrams showing how to arrive at hue/chroma from a point in the RGB cube are drawn in “skeleton” style, showing the edges of the gamut, and the vertices, but not all the colors in between. These are meant to be a concrete geometric picture of the mathematics, and their style adopts the conventions of mathematics diagrams. One particular point is chosen, but symbols are used because it stands in for any point. I added specific numbers, as well, because it is possible to work through the formulae for that point to understand how said formulae operate. So, what would make a good example for these diagrams?

• The numbers should be as simple as possible, so that we can stick as much as possible to easy fractional representations.
• The angle should be large enough so that the portion of the hexagon drawn between the 0° axis and the color in question is clearly visible, and its hexagonal shape comprehensible.
• The color should be somewhat distant from both the origin and the edge, so that the ratio OP/OP′ is clear.
• The point should be somewhere that labels for it will easily fit without running into each-other or other parts of the diagram.
• The angle should be far enough from a multiple of 30° that H and H₂ show some noticeable difference, both numerically and in the diagrams.
• The angle should be far enough from a multiple of 60° that C and C₂ show some noticeable difference, both numerically and in the diagrams.

The diagrams showing comparisons of vertical cross-sections, in the “Lightness” and “Saturation” sections of the article, have a completely different set of goals. These show a full cross section, rather than a skeleton, because their primary goal is comparison of the way these divergent formulae for lightness or saturation act on many different points. Furthermore, unlike a geometrical projection – in which many different points in a 3-dimensional space are mapped onto the same point in a 2-dimensional space – with a cross-section, the color at each point is unique and can be used directly. What would make a good pair of complementary hues for these diagrams?

• The choice should show good separation between I, L, and Y′.
  • Y′ and L are furthest apart for blue/yellow, and identical to each-other for reddish-orange and for bluish-teal. They’re somewhat separated at red/cyan.
  • L and I are furthest apart for multiples of 60°, and identical at points halfway between those.
  • Y′ can differ from L more extremely than I can; therefore, as long we get near to blue/yellow, the three will be clearly separated.
• The hue shouldn’t be an exact multiple of 60°, so we can see some difference between C and C₂ for the diagram about HSI saturation.
• The hue might as well be an even multiple of 10° for simplicity/clarity: we want readers to immediately recognize that the complementary hues are 180° apart.

The “unintuitive RGB” diagram has a much simpler goal, and therefore admits many possible choices of color. Basically, we want an operation (in this case, cutting perceived chroma in half) which is visually quite simple, but numerically confusing when expressed in terms of RGB. The operation could just as well be reducing lightness or changing hue. I tried about 10 possibilities, and the numbers here for orange worked out pretty well.

There are several other example colors used in the article, with still different goals:

An image of a fire breather is used as a graphical demonstration of the divergent definitions of lightness (this was originally in the section about lightness, but was moved by SharkD to the “Disadvantages” section because it is referred to in both places – I don’t really care enough to move it back). For this image, the goal is to have the 5 versions of lightness look as distinct as possible. I tried about a dozen images, which I found by looking through nominees for “wikipedia picture of the year” contests. My criteria ended up being:

• Make sure most/much of the image consists of bright colors, because for neutrals 4 of the definitions agree precisely and the fifth isn’t that far off. Bright colors will completely blow detail out of V
• Try to get some yellows in, because bright yellow is defined to be very light in L* and Y′, and much dimmer in I and L.
• Try to get a few different hues, so we can compare the divergent effects on them.
• Aim for in image with high global contrast, so that light and dark areas can be compared.

An image of a sea turtle is used to graphically demonstrate the kind of effect a change in HSL/HSV hue will have. Again, bright colors were one criterion, but here I also sought an image whose perceived lightness would change quite a bit with a change in hue, and which would still look somewhat realistic after a drastic hue change.

In the “Disadvantages” section, the blue primary is compared with white. The criterion here is to pick a color with V = 1, but with perceived lightness as far as possible from white. A comparison is also made between bright green and pale yellow; the criterion is to pick two colors with S_HSL = 1 but vastly different perceived saturation/chroma. Finally, a comparison is made between a pair of hue rotations of blue/teal colors. Here the goal was to find starting colors for which the same operation clearly had opposite effects.

In the beginning of the article, red, yellow, green, cyan, blue, magenta, white, and black are used as examples. The criterion is hopefully clear.

Okay, so those were my criteria. What was SharkD’s objection? Basically, having more than one example might possibly be confusing to someone. The claim is that a purely hypothetical reader will be unable to immediately realize that two color that look completely different, and are clearly designated by differing numerical color-attribute values, are in fact different colors. This confusion will so overwhelm this imagined reader that he can no longer comprehend any of the text or formulae, and he will decide therefore to give up on the article altogether.

My guess is that this is not the real reason. Instead, my impression is that having more than one example somehow offends SharkD’s sense of order, and that he hasn’t considered that there might be good reasons for using multiple examples; in the absence of such reasons, if the choice of example colors is completely arbitrary, it seems perfectly logical to stick to one example. Hopefully spelling out those reasons in great detail here will help us to understand each-other; unfortunately shorter explanations have, as I mentioned, been ignored.

What is his proposed solution? Pick some single example and then use it consistently throughout the article. Has an example color/hue been suggested? No. Will one example satisfy all of the criteria for different diagrams? Probably not. Who should come up with this example? It’s anyone’s guess.

Would making diagrams slightly less clear for their specific purposes, in the interests of having a single unifying example color throughout the article, actually aid comprehension? There’s really no way to know, without A/B testing two versions of the article on a few dozen or hundred readers, and then quizzing them for comprehension at the end. What would it take to change all the diagrams to fit some common color choice? Several hours of effort. Who would do it? Presumably me, since no one else volunteered.

In short, this is a change which I am unlikely to carry through, unless there is some really really compelling argument for it. I haven’t heard that argument. Someone else is welcome to try to do the work: my guess is that it would quickly become clear why the current examples were chosen.

Cheers,
jacobolus (t) 05:30, 31 March 2010 (UTC)

Jacobolus, thanks for taking the time to answer. Shark, what's your proposed solution? Do you have other graphics you'd like to see used instead? Specifically, I mean, things we could all look at. Also, on a side note, without passing any judgment on this particular point yet, for the moment I fear that this article has much greater problems with being impenetrable to laymen. — e. ripley\^talk 23:40, 1 April 2010 (UTC)

Well, I don't think these considerations should be given more importance than the possibility that a reader might become lost by the colors repeatedly switching. I don't think the charts will become significantly "broken" by using a single color consistently. I don't have alternate graphics on hand, as they would look mostly the same except with a different color. SharkD  Talk  01:01, 2 April 2010 (UTC)

I would suggest that if you'd like to change them, you should probably provide alternatives. — e. ripley\^talk 01:20, 2 April 2010 (UTC)

I will be gone for the holiday weekend but will make some when I get back. SharkD  Talk  03:01, 2 April 2010 (UTC)

From the opposite point of view, by just using one color sample for the entire article we do a different disservice to the reader in that it gives them less of a feel of how the color coordinates work / feel. I can see using one color over a subsection so that various transformations etc. and be seen against the same color. But over the entire article I don't believe it works unless there was a set of colors (not just one) constantly used through out the article to give a little better breath. For example show both the orange and the teal for all of the coordinate pictures. PaleAqua (talk) 01:21, 2 April 2010 (UTC)

Two colors would not be a bad idea. If you could draw a quick sketch of how this would look it would be great. My view is less "fresh" than two months ago. SharkD  Talk  02:48, 2 April 2010 (UTC)

For the cross-sections, there’s no easy way to show 2 specific colors in the context of their containing hue slices, unless we just show two copies of each slice. For the math “skeleton” drawings, I don’t think adding a second color is actually all that helpful, but feel free to try mocking something up. My primary goal for these “skeleton” diagrams was just to make the formulae clearer. They are as far as I’m concerned, a success so long as someone who reads the text, formulae, and looks at the diagrams understands that hue/chroma are made by projecting a point directly into the plane, and that hue = “hexagonal” angle while chroma = “hexagonal” radial distance. Anyone interested in more examples can see a table of ~20 colors along with the values of all these attributes. Or can look to the “swatch charts” at the end of the article which have many more examples. –jacobolus (t) 03:19, 2 April 2010 (UTC)

As an aside, I think we should leave figures that don't show coordinates such as the turtle and firebreather out of this topic of discussion as I don't believe they show an individual sample and just focus for now on the use of images that show coordinates, spot colors and possibly the cross-sections. PaleAqua (talk) 01:21, 2 April 2010 (UTC)

I don't have any issue with those pictures, either. SharkD  Talk  02:49, 2 April 2010 (UTC)

I mentioned them because they also use example colors, just as the figures or little colorswatch templates do; they’re part of the diversity of colors shown in the article, and the criteria for their selection differ from but overlap with the criteria for other color choices. It’s conceivable that images could be chosen that include specific colors used elsewhere. They’re only included so my essay-of-a-response there would be complete. :-) –jacobolus (t) 03:19, 2 April 2010 (UTC)

Translation please

The first sentence is almost meaningless to me. Admittedly I know nothing about this stuff. But since this is a general interest encyclopedia, I really shouldn't already have to have an understanding of the topic to make out the first sentence. (I apologize in advance to the author of the sentence if this sounds harsh, I don't mean it to.)

HSL and HSV are the two most common cylindrical-coordinate representations of points in an RGB color model, which rearrange the geometry of RGB in an attempt to be more perceptually relevant than the cartesian representation.

Can someone explain this to me colloquially, in plain English? Imagine I'm your brother's high school friend and you're trying to explain this to me. Because, quite frankly, this is almost as meaningless as if you'd just thrown random words together to me, and I am a reasonably intelligent person with a college degree. The second clause of the sentence is particularly egregious I think. — e. ripley\^talk 23:48, 1 April 2010 (UTC)

Attempting a quick and dirty stab at this, borrowing from some of the other pages, and not worrying too much about wiki-formatting and links, etc...

• "HSL (hue, saturation and lightness) and HSV (hue, saturation, and value) are two models designed to represent the additive colors of the RGB color models used by television sets and computer graphics in a way closer to how they are perceived. "Hue" represents the pure color--without tint or shade-- and is arrange in a circle from reds around 0 degrees, greens around 120 degrees, blues around 240 degrees and back to reds at 360. Saturation is a measure of how pure the color is, though HSL and HSV are slightly different in how they measure this. (Leave the differences for more details below, get the gist in the intro). Lightness is represents the tint and shade of the color, from 0 being shaded all the way to black and 1 being tinted all the way to white. Value represents how bright the color is from 0 being shaded completely black to 1 being a tint of a pure color. The HSL and HSV color models can be represented as cylindrical color solids as shown at the right."

Hopefully I didn't mess that up to much. PaleAqua (talk) 01:08, 2 April 2010 (UTC)

Thanks so much for this. It is already more understandable. Of course there are more words here, but! If I can understand what is trying to be communicated, then I can be much more effective at editing the article. Thank you. Other interpretations/explanations are also welcome if anyone cares to try. — e. ripley\^talk 01:27, 2 April 2010 (UTC)

"Cylindrical-coordinate representation" is supposed to refer simply to the fact that the models are arranged in a cylinder. Other representations are possible, including the bicone model in the Tektronix image, and the sphere in Runge and Munsell's models (later on in the article). I think "perceptually relevant" is supposed to mean easier or more intuitive. I think "cartesian representation" refers to the RGB arrangement as a cube, though HSL and HSV cubes are also possible. Of course, there's no way for the average reader to know all this. SharkD  Talk  01:31, 2 April 2010 (UTC)

See also the basic idea section that Jacobolus added to the article which is probably more like what the introduction should be. Sort of interesting, how close my explanation above was to the that section considering I didn't look at that section while writing it, though I've read it in the past. Perhaps start the article with the content of the basic idea section, and rename (maybe "overview"?) the current start of the article into a new section just below that? PaleAqua (talk) 02:51, 2 April 2010 (UTC)

I only recently looked at this article, but I gave up reading before even finishing the intro. I'm sure the intro is appropriate for people knowledgeable in the subject, but it is out of place for Wikipedia's intended audience. Changes regarding images and technical wording are meaningless if the intended audience cannot even make it thru the intro, so I think that current efforts to improve the article would be best focused on the intro. VMS Mosaic (talk) 03:14, 2 April 2010 (UTC)

I’m not at all attached to the current intro wording, or to the exact organization of material in the intro, “basic idea”, or “motivation” sections. Which particular words/concepts seem too hard? “cylindrical coordinates”? “rearranged geometry”? “cartesian”? “RGB color”? I’ll take a stab at some other simplifications in the next day or two (for the moment, gotta go), but please, suggest away. I’ve also thought the intro is currently too long; ideas for parts to cut or postpone also welcome. –jacobolus (t) 03:28, 2 April 2010 (UTC)

I think, Jacobolus, that the basic problem is that it introduces far too many concepts that it doesn't explain; for something like this, I don't think it's adequate to just stick in a wikilink and expect people to click 5 times to figure it out (and for "cartesian" and "RGB" there's no attempt to explain relevance at all, not even a wikilink). We need to explain these concepts in a way that makes them relevant to the article and also easily-understood, immediately, where the intro is concerned. Of course these items can and should be examined in greater depth later, but the intro really needs to be more accessible. I realize this is somewhat of an abstract comment, but the intro should explain what this stuff is. I actually rather like PaleAqua's attempt above, with some modifications. — e. ripley\^talk 11:40, 2 April 2010 (UTC)

Yes, I definitely agree with you. It’s a problem that there’s this long dense intro before the summary in language that a non-technical newcomer will understand. The questions are: how much do we actually need to explain in the intro, how much context does it need to provide in itself, what level of concept can we assume of the reader, how much can we fudge/simplify before statements start to be inaccurate, etc. –jacobolus (t) 17:33, 2 April 2010 (UTC)

In the first sentence, what does "rearrange the geometry of RGB in an attempt to be more perceptually relevant than the cartesian representation" mean? I think I know what it means, but the average reader does not expect that level of "deepness/denseness" in the first sentence. While accurate and well written, it is little too much. VMS Mosaic (talk) 03:57, 2 April 2010 (UTC)

It means “apply some math to the RGB numbers (which otherwise form a cube) to make a new set of attributes (arranged in a cylinder) which are more relevant to the way people see and usually think about color”. You’re right though that it should be in plainer language. –jacobolus (t) 17:33, 2 April 2010 (UTC)

For comparison, see the intro for the earlier revision of the article. IMO, the earlier version of the article was easier to understand and should have been kept, and added to, instead of deleted and replaced. SharkD  Talk  04:33, 2 April 2010 (UTC)

I'll take a look at it now. — e. ripley\^talk 11:40, 2 April 2010 (UTC)

Okay. I wrote that version too, in October 2007 (some of the content of those edits came from the HSV article, but the intro section was pretty much new text). But neither one is probably as clear as it should be. Keep pushing us on it, E, until we have something you think is accessible to the average interested high school student (or whatever). –jacobolus (t) 18:54, 2 April 2010 (UTC)