Talk:List of palettes

From Wikipedia, the free encyclopedia

This article is supported by the Color WikiProject, a project that provides a central approach to Color-related subjects on Wikipedia. Help us improve articles to good and 1.0 standards; visit the wikiproject page for more details.
B This article has been rated as B-Class on the quality scale.
Mid This article has been rated as mid-importance on the importance scale.

Contents

[edit] Older discussion

Should Advanced Graphics Architecture for the Amiga be listed at 24 bits (the maximum number of colour precision it uses) or at 18 bits (the maximum number of colours it can display at once)? JIP | Talk 12:08, 15 June 2006 (UTC)

24. —Psychonaut 14:46, 15 June 2006 (UTC)

What about the Windows 640x480 16 color palette?

It's the RGBI palette (legacy from CGA, EGA and VGA). Added mention under the RGBI section. --Viznut 11:50, 21 June 2006 (UTC)

I'm confused. First it says a 2-bit grayscale palette was used on the original Game Boy, then it says it used a 4-color green scale palette. Also, what about other important palettes such as the "web-safe," the 216 color Windows, and the 256 color Macintosh palettes? 69.233.59.38 05:59, 8 October 2006 (UTC)

The introduction of the article states that it deals with hardware palettes, not software-only "web palettes". —Psychonaut 22:29, 8 October 2006 (UTC)
If a device uses 2-bit color, it will have a 4-color palette. Two bits means it has a two-digits in binary, so the colors choices of a 2-bit color palette would be "00," "01," "10," and "11." A bit can be either "1" or "0." Two bits means there are two of them. Two bits would give a total of four different combinations. Concerning the "web-safe" palettes, I am surprised they are not mentioned, especially the 256-color palette, which would be 8-bit color. Why would it mention strange palettes like 9-bit and 18-bit, but not metion the very common 8-bit? Does anyone know this? Jecowa 21:30, 9 October 2006 (UTC)

[edit] Apple IIe, lo-res/hi-res

The Apple IIe section is slightly inaccurate. Those 16 colors were only available in "lo-res" mode, which was 40x40 pixels on the screen, and thus completely incapable of rendering the sample image (which is much larger than the resolution would support.) In "hi-res" mode (280x192), there were only 8 colors, which included two blacks and two whites. The other four colors were created by alternating black and white vertical stripes. The first four colors: black1, green, magenta, and white1, could not be used next to the second four colors: black2, blue, orange, and white2, without creating interference patterns. A "super-hi-res" mode was developped later, which had the resolution of hi-res with the color-palette of lo-res, but required either a hardware expansion card or some special software package to be displayed, I forget which. (There was also a "super-lo-res", which I believe was the same palette as regular lo-res, but with 80x80 pixels.) Neither of these was used much, though, because it was soon afterwards that the newer computer systems came out with different graphics boards in them. (IIgs, I think) I think some of this info needs to be mentionned on the page somehow. (I'd put something up myself, but I don't know how to go about converting that parrot image into the proper color palette, particularly taking the interference patterns into consideration.) --Lurlock 15:59, 21 June 2006 (UTC)

This caveat was already noted near the beginning of the article. —Psychonaut 12:10, 22 June 2006 (UTC)
My point was that lo-res and hi-res modes were the two standards on the Apple II series, and did not require any special tweaking. The image and palette represented there could not be displayed except in super-hi-res, which does require tweaking, so it's not really an accurate represntation of the available palette on the system. It my be worth noting to have a 4-color "hi-res" version shown, with the white/black/orange/blue palette. (Which would probably be more likely than the white/black/green/magenta set, although perhaps both should be shown for the sake of completeness.) --Lurlock 19:29, 22 June 2006 (UTC)
I understand your point; it is already addressed in the article. —Psychonaut 13:20, 24 June 2006 (UTC)
I think it'd be a great idea to have a set of converted pictures showing more or less accurately what the parrot picture would look like in various "vintage" graphics modes. However, I think it is pointless to present a trivial palette-matching conversion in the context of a specific piece of hardware which has severe restrictions in the pixel color assignment. I'm also afraid that such a picture can be quite misinformative despite the disclaimer in the beginning of the article, especially when the reader wants to do eyewise comparisons between the pictures. For example, it is very easy to conclude from the picture that TED is far more superior to VIC-II in showing graphics than it actually is. This problem would be far greater if we had a similarly converted picture using the TIA palette (from the Atari 2600) that has many colors but otherwise very archaic ways of handling graphics. --Viznut 13:32, 7 July 2006 (UTC)
Yes, this is a great idea, but not for this article. It should probably go in some article on archaic ways of handling graphics, not in an article on palettes. I'd like to see this as well—perhaps you could start the article? Gathering the specimens might be a problem, though, for various reasons. For example, different video chips support different screen resolutions, so you couldn't use the exact same test image. Also, doing the actual rendering would be tough, as it would require either manually writing a rendering routine based on old hardware specs, or transferring the test image to vintage hardware or to an emulator and then capturing the rendered image somehow. —Psychonaut 22:55, 7 July 2006 (UTC)

[edit] Some palette maps

I'm collecting, over time, detailed information on computer colour palettes and making maps of them. I hope they will be useful in serving as one-stop reference points on the palettes. When done, I upload the palette map (always a PNG - nothing else can be suitable) onto the Commons. Here they are so far:

  • IBM16 - the CGA/EGA RGBI palette (IBM16 is the shorthand I've coined for that fingerful).
  • Win16 - the Microsoft Windows VGA palette.
  • VIC-II - the Commodore 64 VIC-II palette, based on calculations by Philip "Pepto" Timmermann.

More to be added as I find enough details. I'm currently trying to find about the Sinclair Spectrum palette. As with the C64's, various Spectrum emulators have different RGB values for its entries, so it's in need of research for accuracy (as Timmermann has done for the C64). --Shlomi Tal ☜ 20:57, 26 June 2006 (UTC)


[edit] Confussion about DAC RGB combinations and pixel depth bits

The idea of your article is good, but I think it is not well realized. Palette systems have two magnitudes, often imposed by the hardware available: the number of the RGB combinations that DAC (Digital to Analog Converter) circuitry can offer (usually, due the number of bits employed in each channel) by the graphic device and the number of different indexes (the size of the palette) that can be simultaneously displayed for each pixel on screen memory (dependent of the size of available display RAM in the device).

So, the list of palettes by hardware DAC combinations will be:

  • Monochrome 1-bit, over a white/green/amber-on-black TCR display monitor, or black-on-yellowsh/blue-on-creme backligkted LCD display. Examples: MDA, Hercules.


  • Grayscale 2-bit, four shades of grey. Example: original Gameboy.
  • Grayscale 4-bits, sixteen shades of grey. Example: EGA/VGA text modes and EGA graphic mode with monochromatic TRC display. (In strict sense, RGB 16-color palette viewed without chromatic signal).
  • Grayscale 6-bits, 64 shades of grey. Example: VGA graphic mode with monochromatic TRC display. (In strict sense, RGB 64 of 256-color palette with identical levels for R,G and B values viewed without chromatic signal).
  • Grayscale 8-bits, 256 shades of grey. Example: VGA graphic mode with monochromatic TRC display. (In strict sense, RGB 256-color palette with selected levels for R,G and B values viewed without chromatic signal); TIFF & JPEG grayscale image files.


  • Especific hardware-implemented 4, 16 or more colors palette: not bit nor level combinations of RGB primaries, but fixed ROM/circuitry colors selected by the manufacturer. When see on TV devices, the perception of the colors may be not the correspondent with the value levels for the primaries RGB employed (most noticiable with NTSC TV color system). Examples: Apple IIe, Commodore VIC-20 and C64, MSX systems, Gameboy Color tricky startup palettes for colorize older original Gameboy monochrome games.


  • RG 2 bits+I bit+RGB 1bit background, giving the combinations: red, green, yellow (both low and high intensity) plus a selectable one of black, blue, red, magenta, green, cyan, yellow or white background. Example: palette #2 of CGA graphic mode.
  • MC 2 bits+I bit+RGB 1bit background, giving the combinations: magenta, cyan, white (i.e., the former plus the blue signal), also with two intensities and selectable one-of-eight background. Example: palette #1 of CGA graphic mode.
  • RGB 3 bits, giving the eight (2³ = 8) basic RGB combinations. Examples: Teletext, Videotext.
  • RGB×I 4 bits, giving the former in two multiplyng intensities (i.e., two intensities for each except for black, 0×1 = 0×2 = 0). Example: Sinclair ZX Spectrum.
  • RGB+I 4 bits, giving eight basic RGB combinations in two additive intensities (i.e., the black also has two intensities, 0+0 = 0 pure black and 0+8 = 8 dark grey). Example: CGA/VGA text mode. Note: original IBM hardware and 100% compatible have specific circuitry to vary the dark yellow to brown, but it is internally encoded as dark yellow, binary 0110).
  • RGB 3 levels (not bits), with 3³ = 27 colors. Example: Amstrad CPC464/664.
  • RGB 6-bits, two bits (four levels) for every primary, 4³ = 64 combinations. Example: EGA graphic mode.
  • RGB 9-bits, three bits (eight levels) for every primary, 8³ = 512 combinations. Example: Atari ST with color monitor.
  • RGB 12-bits, four bits (sixteen levels) for every primary, 16³ = 4096 combinations. Example: Commodore Amiga.
  • RGB 15-bits, five bits (thirty two levels) for every primary, 32³ = 32,768 combinations. Example: VESA S-VGA cards 5:5:5 HiColor modes, Targa 16 bits (15 RGB bits + 1 alpha bit).
  • RGB 16-bits, five bits (thirty two levels) for red and blue primaries plus six bits (sixty four levels) for the green primary, 32×64×32 = 65,536 combinations. Example: VESA S-VGA cards 5:6:5 HiColor.
  • RGB 18-bits, six bits (sixty four levels) for every primary, 64³ = 262,144 combinations. Example: VGA graphic mode.
  • RGB 24-bits, eight bits (256 levels) for every primary, 256³ = 16,777,216 combinations. Example: AT&T Targa and Vista, modern true-color display hardware.
  • RGB 48-bits, sixteen bits (65,536 levels) for every primary, 65,536³ = 281,474,976,710,656 combinations. Example: Crossfield and Hell image processing machines.

The list of palettes by size (i.e, by pixel bit depth, independent of the total color repertoire) will be:

  • 1 bits-per-pixel (bpp), two possible color indexes, usually B&W. Examples in white-on-black: MDA, Hercules. Examples in black-on-white: original Apple Macintosh, Atari ST with monochrome monitor.
  • 2 bpp: four possible color indexes. Example: CGA graphic mode.
  • 3 bpp: eight possible color indexes. Example: TIFF image files (exotic, but legal).
  • 4 bpp, RGB+I: eight RGB colors in two intensities (i.e., not true indexes but direct mapping bits). Examples: CGA text mode, Sinclair ZX Spectrum.
  • 4 bpp: up to sixteen possible color indexes. Examples: Commodore 64, Amstrad CPC, EGA/VGA basic graphic mode, Atari ST.
  • 5 bpp: up to thirty two possible color indexes. Example: Commodore Amiga.
  • 6 bpp: up to sixty four possible color indexes. Example: TIFF files (exotic, but legal).
  • 7 bpp: up to 128 possible color indexes. Example: TIFF files (exotic, but legal).
  • 8 bpp: up to 256 possible color indexes. Examples: VGA/SVGA graphic mode.
  • 16 bpp: up to 65,536 possible color indexes. Examples: TIFF and IPTC IIM image files.

Obviously, beyond 16 bpp, direct RGB values (true color) in every pixel is the preferred technique.

Note: the bits in every pixel can be lineally arranged (256 indexes) or be formed by superimposing monochrome independent bit planes (EGA/VGA, Amiga).

Also, we can cite some RGB level arragements commomly used with the 8 bpp palettes as "master palettes" or "universal palettes" (i.e., palettes for multipursose uses):

  • Grayscale 256 levels, doing every triplet primaries having equal values. Used to display pure grayscale TIFF or JPEG images, for example.


  • RGB 6-6-6 levels, 216 combinations. The index can be addresed by (36*R)+(6*G)+B, with all R,G and B values in a range from 0 to 5. Intended as homogeneous RGB cube, it gives true grays. Example: Web palette, Windows palette (this also contain fixed system colors and other intemediate levels of gray, so RGB simple direct addressing is not possible).
  • RGB 6-7-6 levels, 252 combinations. The index can be addresed by (42*R)+(6*G)+B, with R and B values in a range from 0 to 5 and G in a range from 0 to 6. The same case as the former, but with an added level of green due to the greater sensibility of the normal human eye to this frecuency. It does not provide true grays, but remaining indexes usually are filled with four intermediate grays.
  • RGB 6-8-5 levels, 240 combinations. The index can be addresed by (40*R)+(5*G)+B, with R ranging from 0 to 5, G from 0 to 7 and B from 0 to 4. Levels are choosen in function of sensibility of the normal human eye to every primary color. Also, it does not provide true grays. Remaining indexes can be filled with sixteen intermediate grays.


  • Adaptative palette: those whose whole number of indexes are filled with RGB combinations selected from the statistical order of apparience (usually balanced) of a concrete full color original image. There exists many algorithms to select the colors; one well known is the Heckbert's median-cut algorithm.
  • Adaptative palette with transparency: as the former, but reserving one index, usually the number 0, as a transparency mask.

For every of this, gamma correction can be added to the RGB triplets in order to approximate "logical" RGB colors to "physical" ones, especially in TRC monitors.

In addition, I think that the parrot images should not be used as examples of DAC combiantion posibilities, only the templates (because paletted video hardware can not simultaneously display all the DAC combinations, for example the 262,144 of the VGA).

Also, the images in CGA high intensity modes should have the background in pure black, not dark gray, due to the background color are not affected by the intensity attribute.

Finally, the images that shows fixed palettes present in old machines (Commodore VIC, C64, ZX Spectrum) should be adapted to the color blocking (usually 8×8) in which this machines can display colors (no more than two per block).


Yours,

Ricardo Cancho Niemietz, Proffesional programmer and graphic designer of videogames in the last of the 80's-early of the 90's. (You can find my profile on Internet).213.96.157.218 14:14, 21 February 2007 (UTC)

I became a member and I did the most of the changes by myself.Ricardo Cancho Niemietz 11:29, 27 February 2007 (UTC)

[edit] Changes to this page

Hello. I did many changes in this page, including:

  • Splitting the former General section in two hardware palettes and software palettes which clarifies different palette's roles (I hope).
  • Adding of some another cases (Amstrad CPC, MSX, MSX2)
  • Expandig the IBM CGA and the ZX Spectrum sections.

I just became a Wikipedia member to do the changes, due to my interest and background on the matter.

But I think that some work is still pending. There are stub sections (as the nintendo PPU) and some section without a sample image (as the 9-bit RGB section).

With the time, I hope that the original author, I and others will fullfill every known case.Ricardo Cancho Niemietz 17:48, 27 February 2007 (UTC)

[edit] Simulation images

As discussed above, the parot image is used to illustrate the different pallets. On its current state, the article discusses some machine hardware specific palettes and its usage. On that context, I've simulated the parrot as viewed on the Zx Spectrum with it's hardware limitations. Also, there are now other nice simulations, that are very informative. Yet, I think that another image should be used for this, to avoid confusion and to better match the 4:3 aspect ratio of the machines. Also dithering and other techniques should be used to provide a fair comparison. If no one objects, I'll find a better example image and start a example gallery at the bottom of the article. Ricnun 13:31, 6 March 2007 (UTC)

Actually I'm updating this article, and I performed the most of the simulated images, based always in the famous parrot. I agree, this is not the best one to that purpose, but it is usable under the license conditions of the Wikipedia, and are the one which first were in use in the article. I didn't want to remake every image.
But perhaps we can give it the 4:3 aspect ratio putting a 256 level gradient vertical strips at parrot's right side, similar to the moiré patterns found in the ZX Spectrum simulated image. I think this will result more useful, due to the parrot image lacks many hues (it is mainly red and green). I propose to put a gray, red, orange, yellow, green, cyan, sky blue, blue, violet, magenta and purple strips, from black to color to white each. Then we can construct the full palette's "richness" images for every case, and left the actual parrot simulations as they are (because color cuantization will work very bad with this new full color image, specially when down to 4 or 16 colors, I think).
You can email to me through my member page, to send me proofs instead to put them directly in the article, so we'll be synchronized. Together & friends, we can reach the A-class quality degree! Ricardo Cancho Niemietz 19:39, 6 March 2007 (UTC)
Ok, but this will be done time permitting :-) Yes, we can "update" the parrot with the patterns. I did that ZX Spectrum version to try the image modes myself on emulators and TVs. Something along those lines will do, I think, for all the systems. To be as accurate as possible, the master image will have to be 4:3 and of at least 640x480 resolution. But one step at a time, lets work along this lines as possible and we will get there. Ricnun 00:40, 7 March 2007 (UTC)

[edit] NES Palette

The NES palette is available from several sources, but I found this one most readily, thealmightyguru, palette numbers. Does this qualify for inclusion? Inmatarian 02:18, 8 November 2007 (UTC)

I found another source, [1]. I've added a table visual aide, consistent with the other tables of the page using the values from the link I showed last week.Inmatarian 23:25, 15 November 2007 (UTC)

[edit] Article has lots of errors

For example, it classifies the Amiga OCS as "12 bit RGB" with "3 bits per color" which is completely false. The Amiga (along with most machines of that era) was limited to 8 bits per pixel, and thus only 2 bits per color. It achieved its 4096 colors via a "trick" of the hardware (Hold-and-Modify bits).

No error: RGB availability is 12-bit; simultaneous colors are 2, 4, 8, 16 and 32, depending the number of bit planes employed (1 to 5). Extra Half Brite (EHB) and Hold and Modify (HAM) modes can display 64 and 4096 colors at once on screen, via special adjustment. See List of 16-bit computer hardware palettes#Amiga OCS. -Ricardo Cancho Niemietz (talk) 13:00, 12 February 2008 (UTC)

Many other 1970s/80s machines used similar techniques to achieve more colors, but still limiting themselves to an 8-bit byte for each pixel. I think it would be wiser to divide the article by "# of color" not number of bits. - Theaveng (talk) 13:48, 29 November 2007 (UTC)

[edit] RGB palette

When I search for "regular RGB palettes", the only google hits are this article; so it seems to be the product of WP:OR. The term "RGB palette" itself is common, but does not have the meaning that has been used in this article, so that's more WP:OR; these are among the reasons that I'd like to see sources for the concepts around which this material is organized, as well as sources for what the palettes are in the numerous devices and programs discussed. Dicklyon (talk) 17:39, 18 February 2008 (UTC)

There many computer hardware palettes based on RGB exists. In order to make a better compresion on the topic, I splitted the list into "regular" and "non-regular": if every primary R-G-B color has the same number of bits, it is "regular"; otherwise "non-regular". Also, the criteria is well explained and it is not ambiguous. Which is the problem, then? The word "regular"? How many distinct meanings have to you this common word in this context, that you need a citation? Here "regular" is simply the opposed to "non-regular", it is not an academic term (as far I know, there is no academic research on a full listing for the "computer palettes" topic). It is simply a practical criteria. When any wikipedian orders the sections on any article, does he doing "Original Research", or simply giving the article a rational structure? Do you prefer articles without sections? What do you do when you need to rearrange a very large article? And again: follow the links and you'll find all the sources. -Ricardo Cancho Niemietz (talk) 18:48, 18 February 2008 (UTC)
Yes, I can see what you have done. But I don't think it's appropriate. If you look up the term "RBG palette" in books, you find that it is not generally used in the way you've used it; you've made up the meaning that "RGB palette" means a color palette in which subsets of bits are used to independently control R, G, and B. Perfectly logical, but it's not what the term means, and therefore not appropriate for wikipedia to push this new meaning for an old term. If you looke up "regular RGB palette" you get nothing at all; again, perfectly logical, for a result of original research and therefore not appropriate. Maybe your organization is fine if you don't use these terms; or at least explain that you have made up these terms to organize the material, and they are not standard terms. Dicklyon (talk) 19:55, 18 February 2008 (UTC)
Hum, at the end you understood me. The last appears to me the best solution. Although I never tried to impose any terminology (in a field where no standard terminology exists), it seems that the terms I used could be misleaded as authoritative by someone. Even if you found the expression "RGB palette" in various books, the term is always context dependent; it can refer to:
  • the full of the colors that a given system is able to generate[2] (in which case I used the term "full palette"), or
  • the limited selections of colors that a given system is able to display simultaneously from its wider palette[3] (in which case I used the term "selected colors" or "software palettes"), or
  • the hardware registers in which the selected colors values are to be loaded in order to display them[4] (in which case I used the term "hardware palette" or "CLUT"—Color Look-up Table), or
  • a color selection officially standarized by some body (as the well known "Web colors")[5], or
  • the full of the colors a given image file has[6], or
  • the color selection inside a given indexed color image file as GIF[7] (although the expression "color table" is more widely used instead), or
  • even the graphical interface on screen a given software or system provides to a user to select a color.[8]
(The above links were ramdomly picked from your google's search, and they are mere samples of the many different uses of the "RGB palette" expression). So there is no a single "old meaning" and therefore no a "new meaning". I think that perhaps all these definitions should be mentioned in the palette (computing) article, rather than in the listings, but they can be cited there.
The word "palette" applies to all of them in many papers (depending on if it is an engineer's technical spec, an adversitement, a programmers' guide, a image file spec, an user's manual, etc.), and the "RGB" adjetive is used when the colors are produced/encoded by mixing different intensities of primaries red, blue and green. (Ironically, I improved the original 2006 article due to it mixed all these meanings together). I'll try my best to explain these differences in the articles, in order to avoid the impression to be authoritative but merely organizative. Now, I only need to find some time to do it. Yours. -Ricardo Cancho Niemietz (talk) 10:56, 19 February 2008 (UTC)
Except for the GUI one, they all refer to a set of RGB colors; none of them imply by "RGB" that R, G, and B are indexed by separate bit fields, which is what you have used. There is a consistent old meaning, and a new meaning that you've invented that is different. Dicklyon (talk) 15:34, 19 February 2008 (UTC)
In the "Non-regular RGB palettes" section of the List of monochrome and RGB palettes there are samples of not-bits level based RGB palettes, and in the "RGB arrangements" section of the List of software palettes are also another RGB palette arrangements that are not bit level based too. Where it is said that "RGB palette" is a definition of RGB bit fields? This is your own Original Research, not mine! ;-) So I didn't "invented" nothing at all.
OK, so what is it that you mean by RGB palettes, as distinct from other palettes? I'm still trying to understand. Dicklyon (talk) 21:04, 19 February 2008 (UTC)
If I finally understood you (I'm not a native English speaker, only en-2 level, sorry), the problem is that in a generic way, a "RGB palette" is simply a given color selection in which every color is defined by a R-G-B triplet, but in the List of monochrome and RGB palettes article it may be inferred that a "RGB palette" is only a full RGB repertorie, i.e. one in which every possible combination of the R, G, and B components are present giving some level depth for every component. Well, this was no my intention at all. We can agree with the fact that every full "RGB palette", regular and non-regular, are also a "RGB palette"; But, How to name the complete RGB repertories then? There is no stablished terminology at all (the main problem of this issue), so one always have the risk to be accused of doing Original Research. Do you prefer an expression like "full RGB palettes", "complete repertories for RGB palettes", "RGB palettes fully presented" or the like? Why don't you propose an alternative? Meanwhile, I prefer to clarify at maximum in the summary of the article that in the scope of it, the expression "RGB palette" is used as a practical shorthand for "palettes of full combinations of RGB triplets". I'm from and live in Madrid, so we are time shifted. Maybe that when you read this, the changes were made. Yours. -Ricardo Cancho Niemietz (talk) 09:34, 20 February 2008 (UTC)
In other hand, I rewrote the entire palette (computing) article to clarify diverse usages of this concept. As far as I want to fairplay, I left the "need for refs" tag it had. In future second rounds I'll put appropiate references, in order to provide a full consistent set of articles about the computer color palettes topic. By now, you can left your tags, in order to make future improvements to these articles (by me or by another, of course). Yours. -Ricardo Cancho Niemietz (talk) 18:32, 19 February 2008 (UTC)

[edit] still not sure i get it

I tried to define RGB palette to describe what the term means here, but then I hit a snag at the 4-bit RGBI. Is there a better definition? Or is the definition of this article actually suppposed to be "palettes used in hardware" or something like that? Or does my definition only apply to "full RGB palette", of which "regular RGB palette" is a subset but the non-regular ones are not? Too bad all this half-backed original research isn't even clearly defined. Dicklyon (talk) 04:57, 21 March 2008 (UTC)

The summary already states that these are "palettes used in hardware", as you said. So no suposse it, read it! Also, I redefined your definition of "RGB palette" to macth my original intention. Ricardo Cancho Niemietz (talk) 19:34, 27 March 2008 (UTC)