Wikipedia:Featured picture candidates/1 million colors

From Wikipedia, the free encyclopedia

1 million colors[edit]

Original - This image (when viewed in full size, 1000 pixels wide) contains 1 milion pixels, each of a different color. The human eye can distinguish about ten million colors, most of which are outside the gamut of this image. (See image page for algorithm by which image was created.)
RGB channels explained (not for voting)
Reason
No, this is not a joke (you may remember my "single pixel" April joke...) If you think about it, it is really a rather revealing image; every single pixel is of a different color, but still, the human eye is capable of distinguishing ten times as many colors! (Really not practical to create a ten times larger image, cannot be easily viewed in full... So this should fulfill all FPC requirements - even size. ;-)
Tech note: Incredibly, saving this as compressed but lossless PNG, the file size is only 11 Kb - saving in uncompressed TIFF, the same image is a whopping 3 Mb! Saving in "artifacty" JPG would of course change the value of some pixels, at the same time destroying the original intent of the image... Also note that most modern computers can display over 16 million colors (256 levels each of red, green & blue), here only 100 levels of each color are shown.
Articles this image appears in
Color, Color vision
Creator
Janke
  • Support as nominator --Janke | Talk 07:15, 25 September 2008 (UTC)[reply]
  • Support fascinating. DurovaCharge! 07:50, 25 September 2008 (UTC)[reply]
  • Comment Perhaps a bit more commentary on what it actually mean? What are changed along the X- and Y-axis? I can work out the Y-axis but I don't know about the X-axis. --antilivedT | C | G 09:10, 25 September 2008 (UTC)[reply]
OK, good question, answer follows: The value for red changes cyclically in the x-axis, 0-100% ten times in 1% increments, the value for green changes cyclically in the y-axis in the same way, while the value for blue changes by 1% in each "square". Imagine putting all the 100 by 100 pixel squares on top of each other - you would then have a cube, 100 px each side, with R,G and B along the x,y,and z coordinates, each changing 1% for every step. I have also added this info and image showing the R, G and B channels to the image description page. --Janke | Talk 09:44, 25 September 2008 (UTC)[reply]
Why did you use an arbitrary 1 million colours/1000x1000 as the size of the image? This means that it isn't using all 256 values for R, G and B. Also, the statement "The human eye can distinguish ten times as many different colors than are shown here" is a bit confusing/misleading, because it doesn't explain whether that means we should be able to distinguish ten steps between each colour change in this image, or not.. Obviously if you think about it, we can't, because this image is only a very small part of the gamut our eyes are capable of seeing and it is only in this wider gamut that we can fully distinguish that many colours. :-) Colour is such a complex subject and while I commend your attempt to show how RGB colours interact, I'm not sure it it will add enough value.. It may well confuse people more than it educates them! Diliff | (Talk) (Contribs) 10:55, 25 September 2008 (UTC)[reply]
Well, if you check the articles the image is in, there is a reference to the human eye being capable of resolving 10M colors - that sparked my interest, and made me generate this image, to give an idea of what 1M different colors look like. I think the EV is very good in that regard. (I fixed the caption re. "than shown here", thanks for the observation!) Also, this image is intended for those two articles, not to demonstrate the full 24-bit RGB space. A 16-million pixel image would be quite unwieldy (4096 by 4096 px) - this one will display in full on a modern, largish monitor... Hope this explains my reasons. --Janke | Talk 11:09, 25 September 2008 (UTC)[reply]
I don't think you understood what I meant about distinguishing 10 million colours. I know that the human eye can distinguish that many colours, but what I meant was that it cannot distinguish 10 different colours within each 'step' of your image pixels (which is what may be interpreted from the caption) because the gamut of your image is a subset of the gamut our eyes are capable of seeing. To express it numerically, lets say for example that your image contains 20% of the colour gamut that our eyes are capable of seeing. Obviously, this could only be estimated because of course every monitor will have a different gamut and it will also depending on the settings of the monitor. Anyway, if our eyes are capable of seeing 10 million colours within their gamut capacity, that would mean only 2 million colours would be distinguishable within your image (20% of 10 million), no matter how many steps of colour it actually contained. In other words, our eyes could only see twice as many colours in your image, rather than 10 million that our eyes can theoretically distinguish. I know I haven't explained it particularly concisely, but I hope you understand what I mean now. This is why I think the caption is a bit confusing/misleading. Did you change the caption? It still reads the same to me? Diliff | (Talk) (Contribs) 12:05, 25 September 2008 (UTC)[reply]
I understood the first time... ;-) Click "purge cache" if you don't see the changes. I'm very aware of gamuts etc, working in the film&video business. I just felt it's too much info to put in a caption - explaining that most of the 10M distinguishable colors lie outside this image, so to speak... That's why I chose 1% steps - to give an idea to the casual reader of the color articles of what 1M colors look like. And, most of these should be distinguishable! (Maybe not all in the top percentages of blue...) I furthermore added "most of which are outside the gamut of this image" to the caption. Satisfied? ;-) --Janke | Talk 12:31, 25 September 2008 (UTC)[reply]
I'll Weak Support for its enc value, but I think a better choice of illustrating colours would be to use a more natural colour space, like HSL or CIELAB. --antilivedT | C | G 09:23, 1 October 2008 (UTC)[reply]
  • Support. Really great and original. Good choice! Intothewoods29 (talk) 17:23, 25 September 2008 (UTC)[reply]
  • Comment Interesting. Would it be better to make the image so each individual color is something like a 2x2 or 5x5 or 10x10 block at full resolution, so that viewers could appreciate the incremental differences Calliopejen1 (talk) 19:40, 25 September 2008 (UTC)[reply]
Not really, unless you'd want to load a 10,000 px wide image... but you can zoom into the image by loading it into photo software. --Janke | Talk 20:12, 25 September 2008 (UTC)[reply]
  • Comment Maybe I get this wrong, but I think this picture does not shpw 1 million different colours but only 100000. The pattern is repeated on the x-axis ten times, or do I miss something? 91.63.108.110 (talk) 21:14, 26 September 2008 (UTC)[reply]
    • Oh my god, I cannot believe no one noticed that before. Thanks for the great joke, Janke! (because you cannot be serious now, can you?) --Dschwen 14:35, 27 September 2008 (UTC)[reply]
      • The blue value seems to depend on both x and y. Anyhow, if you guys think this thing is feature worthy, go ahead. This was probably my last visit here for now. --Dschwen 15:34, 27 September 2008 (UTC)[reply]
        • Yes, the blue changes by 1% in every square, so there are indeed 1 million colors in the image, no joke. --Janke | Talk 15:02, 28 September 2008 (UTC)[reply]
        • That was a bit of a premature overreaction you had there Dschwen... All you have to do is look at the blue channel in the nomination to see it does change along the Y axis as well as the X. An apology to Janke, perhaps? ;-) Diliff | (Talk) (Contribs) 09:39, 1 October 2008 (UTC)[reply]
          • Heh, yes. Sorry Janke, you deserve better than my unfiltered bursts of frustration. In my defense: I spotted my mistake an hour later before any other comments were made. I could have chose to remove it entirely, but instead just struck it. So, it looks like this is getting promoted with 6:1 votes. Congratulations! Maybe I nominate the Gimp-Color-Chooser dialog, or the test image of our local TV station. What, copyright violation you say? Don't worry, below the threshold of originality... --Dschwen 20:06, 2 October 2008 (UTC)[reply]
  • Support. Fascinating and informative. Mostlyharmless (talk) 00:43, 2 October 2008 (UTC)[reply]
  • Oppose This isn't the ideal way to view a colour gamut, as the perception of a colour is greatly affected by the surrounding colours, so you can't really just pack them all together like this. See this optical illusion for an example of the problem. —Pengo 04:15, 2 October 2008 (UTC)[reply]
Actually, this isn't intended to illustrate a gamut (even though I was persuaded to include the word in the caption) - the intention is to give a "compact" idea of 1 M colors, in the articles it's in - as you may notice, I haven't included it in RGB or gamut articles, only in color and color vision... --Janke | Talk 06:54, 2 October 2008 (UTC)[reply]
I agree with Janke on this, the only reason gamut should be mentioned is to clarify what this picture shows, and to avoid confusing people into thinking that a human could see all 10 million plus colours within the gamut of this image. Then again, someone that doesn't have a good understanding of colour in the first place would likely be confused by many of the concepts of colour, so it is a difficult case. Diliff | (Talk) (Contribs) 07:38, 2 October 2008 (UTC)[reply]

Weak support. Does do what it intends to do, but apart from being an interesting pattern of discrete colours, I do think it has somewhat limited educational value. Diliff | (Talk) (Contribs) 07:38, 2 October 2008 (UTC)[reply]

Promoted Image:1Mcolors.png MER-C 07:38, 3 October 2008 (UTC)[reply]