Talk:Gamut

From Wikipedia, the free encyclopedia

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

Contents 1 Digital cameras 2 Relative Sizes of Gamuts 3 Gamut and syntheticable colors 4 Color gamut as percentage to NTSC 5 Is this relevant? 6 The "Y" in the gamut image 7 Avoiding unintentional puns. 8 Limitations of color representation 9 Discretisation of gamut 10 Gamut trivia?

[edit] Digital cameras

Gutza, your contribution about digital cameras and dynamic range looks like a valuable information for people interested in photography, but I'm not convinced that it actually belongs in a page about "gamut".

I'm not that much of an expert in photography terminology (I do know a bit about how cameras and CCDs work, though), but I think your discussion about "dynamic range" is about the range between the black point and the white point. Gamut refers to the shape of the area perpendendicular to that line. I agree with you that megapixel counts on digicams are misleading since typically 1/2 of the pixels represents green, and 1/4 red and blue, but that has absolutely nothing to do with the color reproduction. You mention uniformity in recording all colors in the spectrum at a given brightness.. Don't know what you mean, but gamut refers to how colors are reproduced, not whether they are recorded. If you photograph violet light (400 nm) with a film, the print will not look violet. Also, different brands/types of film have different behaviour in color reproduction, some being better for human skin colors, others better for use under fluorescent lighting. That implies that different films have different gamuts.

Your text is also a bit non-NPOV, with words such as "honest", "cheat", and "satisfy sale characteristics".

Han-Kwang (talk) 12:45, 20 May 2004 (UTC)

Hi Hankwang,

Sorry for the way-late reply, see my user page for details. For an explanation of my apparent POV, please see this page. Both streaming video encoding systems and digital camera manufacturers "cheat" with their products by encoding as little red and blue information in their resulting output as necessary to be unnoticeable. Film doesn't do this. Check it out with your own digital camera, take a random photo, zoom in to 200% and look at each discrete channel if your software supports this -- you'll see what I mean. This is not meant to be POV, it's the only wording I could find -- and they are cheating, because no camera manual will tell you this, yet they all do it and everybody in the business knows they're doing it. What would you call that? Anyway, if you find a milder wording which does convey the meaning of what I'm babbling in this paragraph, by all means, be bold! :) --Gutza 12:04, 30 Aug 2004 (UTC)

P.S. You might be right about the dynamic range issue, I'll try to look into that if I have the time. Please do fix it if you know for a fact that only brightness is what is meant by dynamic tange in film. --Gutza 12:10, 30 Aug 2004 (UTC)

[edit] Relative Sizes of Gamuts

I'd like to see some references for the claims made in the article about which types of display systems have the largest gamuts. I not convinced that the gamut for a laser based system would be larger than that of a CRT. It may be, but it seems to me that unless a number of lasers of different colors are used, the problem with a tringular shaped gamut would still remain.

I wasn't too happy with the comparisons either, but I will observe that the area which can be covered by a triangular gamut varies depending on two factors: The saturation of the three primaries chosen (so, distance from the central white on the chromaticity chart) and the particular wavelength of the three primaries (so, positions along the edge of the chromaticity chart). In theory, a laser-based three-primary system should be able to achieve the largest gamut because, unlike chemical phosphors on a CRT screen, the lasers are always fully-saturated (because they can be truly monochromatic) and the wavelengths can be chosen pretty-much arbitrarily. As far as I know, you can't choose phosphor wavelengths arbitrarily while maintaining full saturation. I don't know how well this theory plays out in practice.

(And, of course, systems with more than three primaries can cover much larger gamuts, but I don't think there's any disagreement over that! :-) )

Atlant 14:19, 10 Mar 2005 (UTC)

I have a feeling that the comparision between CRT and LCD isn't up to date, atleast when it comes to today's LCD screens. I also missed the plasma screens in here, and the projectors that use three light sources and mirrors to project images on a grey or white textile. This taps into a vast field of home entertainment technology, so perhaps a demarkation between the Gamut page and other pages/categories is in order first :-) —The preceding unsigned comment was added by 129.241.252.14 (talk • contribs) .

[edit] Gamut and syntheticable colors

I don't understand all that discussion about gamuts: my eyes have three different sensors for red, green and blue. Therefore, the only thing my TV/LCD screen/etc needs are subpixels with the colors red, green and blue to make me happy (= give me all colors I am able to see).

As i'm limited to these three sensors, it should be of no importance what spectrum my LCD backlight emits, as long as I get red, green and blue colored light beams to my eyes. Also, my CRT/LCD screen should be able to represent all the colors you (=human person) can see in real nature, as (you already suspect it) you can see only a mixture of the colors red, green and blue.

who cares about the adding of an additional cyan subpixel in a TV, as I don't have a sensor in my eyes to exclusively see this color. I have green- and blue-sensors in my eyes to do this job. And therefore cyan should be obsolete, as the green and blue subpixel in the TV can already fake this color.

in a nutshell: please explain, why gamuts and/or gamut enhancements are important - i won't see the additional colors anyways, or will I? A bigger gamut will not give me more colors than a smaller gamut that already covers RGB.

thank you so much! --Abdull 18:49, 19 Mar 2005 (UTC)

I'm not a specialist in CRT/LCD displays, but I will tell you this: a bigger gamut will definitely give you additional colors -- it just depends on what you mean when you say "color". It won't if you mean "primary color", but then you should be happy with a display system which can only render pure red, pure green and pure blue. But you wouldn't be happy with such a display system, because you wouldn't be able to see any distinction between dark red and light pink. If you define "color" the way women do when they go shopping for clothes (and which is the generic definition of the term), then yes, a bigger gamut will certainly give you more colors. --Gutza 01:22, 14 August 2005 (UTC)

Also, if you have specific sensors in your eyes to do that job, you are exceptional -- see Color vision for an explanation of how the eye works! As for CRTs, look at the sample CRT gamut image on the article page -- the gray area represents all of the possible frequencies of visible light, and the triangle the colors the CRT is capable of producing. The part of the gray area that is not covered by the triangle represents frequencies that the CRT is not capable of producing! (As a trivial example, a CRT cannot produce monochromatic red like a laser can.)

Similarly, your eye can detect variations that a four-color printing process cannot produce (but that, say, spot color can), or variations that printing technology cannot produce at all. Gamut is a technological limitation, not biological. Of course, if you do not believe that your eye is capable of calculations of frequency at a higher accuracy than those CRTs or printing processes are capable of reproducing then none of this will convince you, but it remains that your eye is capable of that. — mendel ☎ 04:52, September 8, 2005 (UTC)

The underlying process you're failing to recognise is that the human eye doesn't have "pure" red/green/blue detectors; the wavelength-responses of the detectors overlap significantly. This means that any physically realisable light source (ideally a monochromatic source) *still* cannot excite a single colour-channel in the eye. Even with three optimised monochromatic r/g/b sources you still cannot reach the full gamut of the eye. Real-world phosphors etc fall short even of that optimum... looking at the CIE Colour Matching Functions should help aid your understanding. 80.189.149.141 13:15, 1 April 2006 (UTC) Andrew http://www.techmind.org/colour/

I'd like to chime in with an example: light that is at the wavelength where your red receptors are most sensitive will register as red. But it will also excite the green receptors in your eye to a small extent. If your eye is then exposed to light of a longer wavelength (in the direction of infra-red), the red receptor will still be excited, but the green less so. Therefore, these two red frequencies are seen as different colors. If you have only one source of red light and one of green and want to mimick the proportional excitement of the receptors seen with the two reds, you'd have to use more light and therefore get brighter colors than the true ones. This therefore limits the range of colors available. 129.241.252.14

[edit] Color gamut as percentage to NTSC

I've encountered that some LCD and LCD-TV manufacturers are using persentages to some standard space when describing the color of their panels: [1]

• 100% EBU Color Space - 72% color gamut, EBU Color Coordinate [2]
• CRT-grade color gamut at 72 percent NTSC [3]
• Samsung's 40” LCD TV (LE40M61B) was honoured with an EISA award for its premium display technology in terms of contrast ratio (5000:1), color expression (6.44 billion colors, 92% color gamut for NTSC) , viewing angle (178 °) and response time (8 ms)

I understand that EBU and ITU709-4 are gamuts in the RGB color space which almost exactly copies sRGB, but then I'm confused with references to NTSC gamut. Is there a standard NTSC gamut defined somewhere, or maybe they are talking about CIE RGB (1931)?

PS. Don't bother, I've found a reference to NTSC RGB here [4] and then right on the RGB color space :)

--DmitryKo 19:08, 31 August 2005 (UTC)

According to Poynton (www.poynton.com), the original NTSC used a more saturated green phosphor than any television produced in the past 30 years. The original phosphor fell out of use owing to being too slow (long persistance) and maybe cost. Remnants of the old specification persist in documents and standards, but have little practical relevance these days.

When a colour-space is described as such-and-such percent of NTSC, we're talking of area of the gamut in CIE 1931 x,y space ... which is far from perceptually uniform. Consequently, in my opinion, %NTSC should be consigned to specmanship and marketing blurb, and has no place in technical discussions! 80.189.149.141 13:24, 1 April 2006 (UTC) Andrew http://www.techmind.org/colour/

[edit] Is this relevant?

I'm trying to understand gamut here, so tell me: is this relevant to the concept? It seems to have to do with the capacity to more accurately reflect brightness ranges, and brightness is a component of color. So would the described monitor have a higher gamut than a normal CRT? —Simetrical (talk) 02:34, 6 October 2005 (UTC)

I'm having a bit of a tough time figuring out exactly what that article is talking about, but I think they've just overcomplicated things. Dynamic range in that context is luminance; it's a one-dimensional measurement. If a display can produce darker blacks and brighter whites than another then it has a higher dynamic range.

There's really no concept of a "higher" gamut, because a gamut is a description of exactly which colors a display can capture. You'll notice that the CIE gamut graph on the article only goes from white to pure colors (although the gamut graphed on it doesn't reach the pure colors at the edge of the graph). If one device's gamut is a superset of another's -- that is, it can produce all of the colors that the other one can produce, and then some -- then you could say that its gamut was "better", but usually gains in some colors mean losses in another.

So the dynamic range of a device is the range between the darkest dark and the brightest brite of a display, while the gamut is the range between white and pure color, for all colors, that a device can display. There's more at high dynamic range imaging and high dynamic range rendering but both are in a bit of a state of disrepair.

Does that help any? — mendel ☎ 03:42, 6 October 2005 (UTC)

Oh, also see the section in the article about a 3D gamut; the third dimension is brightness, and that axis measures dynamic range (and therefore a normal 2D gamut graph omits dynamic range). — mendel ☎ 03:55, 6 October 2005 (UTC)

I think I get it a bit better now. So in theory, this monitor would have a larger gamut than a typical LCD screen. I still don't get why we can't get a wider range of color—isn't the only property of a photon frequency/wavelength,

And amplitude, for brightness! (Yes, waves and particles, but you know what I mean.)

and can't we produce a photon of any wavelength?

Not in every medium, no. Gamut details exactly how a particular medium fails to produce or reflect photons of any wavelength. It's an engineering problem, not a physics problem; it's expensive to produce computer monitors that display far more color than a user would typically need, so instead they produce them with smaller gamuts that will suffice.

But what about pigment? Is it just prohibitively expensive to have enough primaries to reproduce a near-perfect gamut in commercial color printing, or is it actually impossible with current technology? Again, I'd imagine pigments with a huge variety of reflective properties can be created. —Simetrical (talk) 04:35, 7 October 2005 (UTC)

Yes, that's exactly it. CMYK has a limited gamut but it's a limited gamut that's satisfactory, so it gets used. When people need a wider gamut they throw more money at the problem and use 6-color or higher separations, or spot color, or something that has a gamut that better meets their needs. Four-color is good enough and cheap so that's what's widespread. 6-color printers are showing up in the home market these days, though. To ask why a device doesn't have a wider gamut is like asking why an engine doesn't produce more horsepower and torque; gamut (or engine output) is one of the things designed into a device and balanced against cost, reliability, and so forth. Of course there are some absolute limitations in particular mediums as well. — mendel ☎ 14:52, 7 October 2005 (UTC)

Hmm. I'd be interested in seeing a side-by-side photo comparison of two monitors with different gamuts showing the same image. Could this be easily simulated in PSP or something? I think it would be useful for the article. —Simetrical (talk) 21:12, 7 October 2005 (UTC)

Not accurately, because you'd be looking at both through a device with its own gamut. An example would be useful but it'd have to be an extreme one, I'll see what I can think of. — mendel ☎ 23:36, 7 October 2005 (UTC)

Of course, but both will have their different gamuts reduced by the same amount in display on the end-user's monitor, preserving the difference between them. Rather like this picture from the article I linked: the HDR monitor is technically being displayed as a sort of gray rather than pure black, just like the normal monitor, but the normal one is grayer—the difference remains very noticeable. —Simetrical (talk) 03:11, 9 October 2005 (UTC)

[edit] The "Y" in the gamut image

http://www.techmind.org/colour/rogues.html explains the effect is an artifact, and presents "fixes". Wikipedia probably can't use his alternatives because of the usual copyright crud, but I can render equivalents (for many devices) and toss them into the public domain if this would be deemed useful for this article. mdf 18:54, 17 February 2006 (UTC)

Sounds good to me (and the image in the article claims to be in the public domain, so you can free any derivative works you create).

Atlant 19:18, 17 February 2006 (UTC)

An initial rendering is now available for comment - see at right. The addition of text to the graphic make it less neutral, but in this case I think it's addition is justified. I also think the dark background makes the gamut itself stick out better than if the background was white. The out-of-gamut region has been rendered as grey for similar reasons. Note that many gamut's can be rendered at will, and a large 3x4 or 4x4 table of gamuts is also possible with minimal effort. mdf 19:14, 22 February 2006 (UTC)

[edit] Avoiding unintentional puns.

"in addition, for subtractive color systems, such as printing, the range of intensity available in the system is for the most part meaningless outside the context of its illumination."

Can this be reworded to avoid using "addition" and "subtractive" in the same sentence? I can't think of anything all that elegant. 69.16.168.94 21:23, 23 February 2006 (UTC)

Sure. "Subtractive", being the appropriate technical term, can't really be changed here, but "in addition" has many synonyms: "moreover", "furthermore", etc. I'll fix it right now. -- Stebulus 21:13, 15 July 2006 (UTC)

[edit] Limitations of color representation

Even with laser sources you need at minimum 5 different wavelengths to cover the gamut of the eye, and ideally 6. The section kinda implies that it is easy to get a good gamut with lasers, but it's not at all. (reference, Hans, Practical Holography)

[edit] Discretisation of gamut

The comparisions between gamuts does not discuss the discretisation of the colour space. RGB for example, is limited to 16777216 colours. While a lot, it can only distinguish 256 shades of grey. <a href="http://r0k.us/graphics/colSpace.html">this page</a> says that we can "probably" distinguish 85 shades of grey, while <a href="http://www.cs.dartmouth.edu/~rockmore/m5s01/MATH5pk/index.html">this page</a> has an example of 256 shades of grey. It seems the border between shades is still detectable, even if they look the same when apart. Also, as <a href="http://www.techmind.org/colour/">this page</a> says, the number of distinguishable colors in a given part of the chromaticity diagram is not constant. This means some discretisations will be better than others. The large part outside the triangle in the chromaticity diagram is such an area of low resolution —The preceding unsigned comment was added by 129.241.252.14 (talk • contribs) .

(Just FYI) "RGB", per se, does not imply 8 bit quantization so only 16777216 discrete colors. That's just one common implementation of RGB encoding. For example, analog RGB obviously is only limited by the signal-to-noise ratio.

Atlant 12:43, 6 September 2006 (UTC)

[edit] Gamut trivia?

The corners of the triangle in the chromaticity diagram are the primary colors used in the RGB display. Because of the shape of the diagram, to be able to represent all visible colours, one would need one or two primaries that were outside the visual range. Paradoxial, wrong logic or a trivial limitation of the RGB approach? —The preceding unsigned comment was added by 129.241.252.14 (talk • contribs) .