Talk:Grayscale

Example pictures

I put up the picture examples, but I couldn't quite figure out what to do with the formatting. If anyone has any better ideas to make the image/caption placement look less clunky, go right ahead. -Branddobbe 05:25, Feb 17, 2004 (UTC)

Hello, I did it. I simply added <br style="clear:both" /> before the image links.Ricardo Cancho Niemietz 16:02, 6 March 2007 (UTC)

Unifying page of representations kinds

I just created the Binary images, and then i thought would be rather to unify the grayscale, binary and full color in a unique page. Well, i just felt a comparison pages is needed (even if the other pages is . Currently this comparison is presented by Grayscale page. Ribamar 05:00, 4 Jul 2004 (UTC)

What? -Branddobbe 06:39, Jul 4, 2004 (UTC)

PseudoGrey

I have described a technique to encode 1786 shades of grey using a 24-bit palette. It is called pseudogrey, and described on the linked page. I know editors are supposed to be bold, but since it is my own page, I am hesitant to add reference to it myself. Several people have adapted the technique as filters for Photoshop and achieved extremely smooth greyscale images. (The pseudogrey page includes references.) So I'll ask another editor to review my page and decide if the information is relevant to the Wikipedia. —The preceding unsigned comment was added by Dozen (talk • contribs) 06:02, 1 January 2007 (UTC).

Simply calculation RGB to grayscale added

Hello. I added a "converting a color to grayscale" section which describes the very simple and common algorithm to convert any color to the correspondent gray level.

Also, I think that the expressions "continuous tone" and "halftone" should be mentioned to clarify when a grayscale image is rendered with one or another techniques ("contiuous tone" is that of black-and-white photographs and "halftone" is that of printed papers). Then, a link to the correspondent article "halftone" can be set.Ricardo Cancho Niemietz 16:48, 6 March 2007 (UTC)

Other RGB to gray conversions?

Trying out Photoshop (PS Elements 3.0) and its "Image, Mode, Grayscale" command on pure RGB swatches, it is using this formula:

50% red + 86% green + 28% blue

Given the conversions of other swatches (e.g., pure yellow becomes 97% white instead of 100%, RGB [50%, 50%, 50%] becomes 50% white instead of 82%, etc.) it appears to be converting to some intermediate color space and then to grayscale.

The overall effect is to make blues come out darker than reds, greens significantly lighter, and harder to tell between greens and yellows. Upon reflection, though, this approach seems to do a better job of converting colors to brightness values; the resulting contrasts look more reasonable.

The conversion formula I provided is a very simplified form of that used to obtain the luminance signal in the RGB to YUV, YIQ and YPbPr/YCbCr color spaces, employed mainly in TV and video (both PAL and NTSC standards, along with composite video). The L*a*b color space, used internally by Photoshop as its interchange model uses a very similar formula. Please, see the articles.

You are wrong in your computations: pure yellow (R=100%, G=100%, B=0%) gives 89% with the formula provided, not 100%: (100×0.3)+(100×0.59)+(0×0.11)=89; middle gray gives 50%, not 82%: (50×0.3)+(50×0.59)+(50×0.11)=50, etc. So the formula works well, as you noticed yourself with Photoshop. As a result, due the different sensibility of the (normal) human eye to the RGB primaries (the greater to the green, the lower to the blue), the correspondent gray to pure blue is darker than that of the pure red and both darker than that of the pure green, etc.

Perhaps your confussion comes from the gamma correction computation. The formula I provided performs the conversion in a linear way (we are talking about video signals), not in a rendering way. The Photoshop settings let you to calibrate the output according to some typical gamma values to render the internal lineal values to exponential ones. The values you see as percentages in the grayscale slider are shown in output values by default. Please use the info window, set the first eyedropper to show RGB values and set a second eyedropper to show grayscale values. Fill an area in an RGB image with the color value (R=128, G=128, B=128) and hover the mouse over it. The info window shows a grayscale value below 50% (the exact value depends on the color calibration settings), but you'll agree that 128 for every primary is the rounded value of the 50% of 255, the maximum possible value. Convert the RGB image to grayscale, and hover the mouse again. You would get the same results in the info Window. So the internal pixel values and output values are not the same, but are preserved during conversion.

We agree in that Photoshop is an excellent image processing application, but it is not a "color laboratory".Ricardo Cancho Niemietz 09:00, 6 June 2007 (UTC)

Where did the grayscale percentages come from?

"30% of the red value plus 59% of that of the green plus 11% of that of the blue", Who discovered these numbers? —Preceding unsigned comment added by 134.129.90.92 (talk) 05:28, 2 September 2007 (UTC)

As I understand it, the coefficients mimic the relative perceptual RGB color sensitivity of the human eye. The values have been widely disseminated since the advent of the NTSC and PAL television standards where they are used to calculate the luminance signal or voltage level that a video camera produces when it "sees" any particular color. Xot (talk) 14:23, 8 June 2010 (UTC)

That’s not quite right (and neither is this article; I’ll try to correct it). Luminance is additive, so adding up the luminance of the three primary light sources used in a color on a computer display yields the luminance of the color. Adding up gamma-corrected R, G, and B values does not, however, yield lightness. See that lightness article for some explanation. –jacobolus (t) 21:44, 8 June 2010 (UTC)

I don't understand this, but this was why i came to this wiki page. I think the conversion formula is very important! Why is it not possible to just add up 1/3(r+g+b)? — Preceding unsigned comment added by 77.176.68.153 (talk) 15:48, 25 July 2011 (UTC)

Because the R, G, and B primaries aren't equally bright, with respect to human perception. The “green” primary is much brighter than the other two, and the “blue primary” is much darker. –jacobolus (t) 16:59, 25 July 2011 (UTC)

I added the original formula for computing the luma component of the YUV/YIQ color models (used by PAL, SECAM, and NTSC video), as well as the newer formula of ATSC (used for HDTV) to the article. These are the popular 30%R + 59%G + 11%B and 21%R + 72%G + 7%B computations. These were determined by the ITU-R. — Loadmaster (talk) 16:09, 26 August 2012 (UTC)

Black and White are not colors, are they?

I think the sentence "images with only the two colors, black, and white" needs amending - as I understand it, black and white are technically not colors, they are levels of light and dark which can be applied to colors, or stand alone as shade gradients. SpaceHistory101 (talk) 16:57, 16 June 2009 (UTC)

Well, technically white is the equal, isotropic combination of all of the colours in the visual spectrum (that is, the rainbow - which is white light diffracted). or, in the case of video screens, of red blue and green in equal proportions. Black is the absence of colour; that is, the absence of light. A black object absorbs all spectra of light, while a white object reflects all spectra of light. Firejuggler86 (talk) 01:02, 3 February 2021 (UTC)

Another "Gray scale"

In shaft encoders and the like, it is possible to sequence the codes returned for different positions so that on every transition from a position to an adjacent position, only one digit changes in the binary code for the positions. This is also, as I remember it, called a "Gray scale", named for Mr Gray. See...

http://en.wiki.x.io/wiki/Gray_code

.. so it IS in Wikipedia, but I think many users will look under "Gray scale", and not find "Gray code"

Perhaps a "see also" is needed to someplace else? And a "were you looking for? at the head of the article?

Tkbwik (talk) 13:10, 10 February 2010 (UTC)

Bits per sample

The article states: "Sixteen bits per sample (65,536 levels) is a convenient choice for such uses, as computers manage 16-bit words efficiently. The TIFF and the PNG (among other) image file formats supports 16-bit grayscale natively". When I look into the TIFF specification at http://partners.adobe.com/public/developer/en/tiff/TIFF6.pdf I find in section 4 Gray scale images: "Allowable values for Baseline TIFF grayscale images are 4 and 8, allowing either 16 or 256 distinct shades of gray" So I think the article is wrong about TIFF supporting 16-bit grayscale natively. An explanation or confirmation (or correction of the article) would be appreciated. Vollklau (talk) 12:35, 20 April 2011 (UTC)

I don't think that's right. The same document you link says to use (8, 8, 8) as the value for RGB images. I'm pretty sure TIFF allows setting bits/sample to a variety of other values in either grayscale or color images (for example, I can save a 16 bit/pixel grayscale TIFF from Photoshop, and it seems to work just fine in other software). –jacobolus (t) 20:31, 20 April 2011 (UTC)

Photoshop implements a lot more than the "baseline" TIFF. I'm not sure the baseline is that useful a concept. Nor is the "natively" in the article very meaningful. It should probably be removed. Dicklyon (talk) 04:21, 21 April 2011 (UTC)

additional citations

I'm adding several citations for key aspects of this article and removing the Refimprove ("This article needs additional citations for verification...") box from November 2008. DavRosen (talk) 19:22, 14 October 2013 (UTC)

Should this page be called 'Grayscale image' ?

does 'grayscale' refer to the data itself. (like RGB image and image aren't the same page.). "a grayscale is a scalar intensity (whatever)", .. there could be an anchor in here defining 'grayscale value'

• I think it should be renamed grayscale image. Grayscale is an adjective, and wikipedia articles should preferably be named with nouns. --Jorge Stolfi (talk) 10:31, 23 November 2021 (UTC)