Talk:Bayer filter

From Wikipedia, the free encyclopedia
Jump to: navigation, search
          This article is of interest to the following WikiProjects:
WikiProject Color (Rated C-class, Mid-importance)
WikiProject icon This article is supported by WikiProject Color, 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.
C-Class article C  This article has been rated as C-Class on the project's quality scale.
 Mid  This article has been rated as Mid-importance on the project's importance scale.
WikiProject Film (Rated C-class)
WikiProject icon This article is within the scope of WikiProject Film. If you would like to participate, please visit the project page, where you can join the discussion and see lists of open tasks and regional and topical task forces. To use this banner, please refer to the documentation. To improve this article, please refer to the guidelines.
C-Class article C  This article has been rated as C-Class on the project's quality scale.
Checklist icon
Taskforce icon
This article is supported by the Filmmaking task force.

Merge between Bayer filter and Bayer sensor[edit]

proposed by user:Abdull the 1st mrch 2006

  • agree
  1. to Bayer filter, bayer sensor is just image sensor + bayer filter. Plus Special:Whatlinkshere/Bayer_filter wins hands down versus Special:Whatlinkshere/Bayer_sensor. There is too many articles on the same subject, see also Demosaicing; RAW image format; Mosaic (digital image); Calculating RGB values... --Marc Lacoste 14:13, 1 March 2006 (UTC)
  2. It seems like a good idea, much of the information is duplicated. Henrik 23:23, 7 March 2006 (UTC)

Done --Marc Lacoste 22:46, 19 March 2006 (UTC)

Excess green.[edit]

Does anyone know if the 50% green sensor presence is the reason why most people who work with imaging always complain about digital cameras and excessive green? nihil 21:35, 16 August 2007 (UTC)

I would disagree with the premise, to start with, but no. Dicklyon 01:23, 17 August 2007 (UTC)
Well, I've worked with several people involved with desktop publishing, digital imaging and such things. Most of them would start treating images by correcting colour balance, subtracting some green, and then bragging something like "Pff... digital photography, always too much green!". nihil (talk) 19:19, 17 December 2007 (UTC)
I read somewhere (sorry no ref...) that CMYK bayer filters are sometimes used primarily to provide more accurate colour separation for CMYK printing. Interestingly, because the CMYK filter dyes are "paler" than RGB, CMYK filters have higher transmittance than RGB, making them slightly more sensitive. (talk) 12:27, 15 January 2009 (UTC)
The 1976 Bayer Patent illustration shows a YCC (YCbCr?) filter rather than RGB. If my guess is correct, the Luma (Y) sites (requiring no green dye) would be more sensitive than the corresponding Green sites on an RGB filter. If so, the YCbCr colour demosaicing algorithm would need to take this into account to avoid an "excess green" caste. (talk) 12:27, 15 January 2009 (UTC)
@Nihil, such claims about digital photography being too green come down to either personal preference or an incorrect colour temperature setting in the camera (or even a mis-calibrated monitor, if the 'too-green' complaints are always coming from the same person). It is not related to the area of the sensor covered by green filters being greater than that for red or blue. Due to colour correction inside the camera (colour balance as well as primaries adjustment), the amount of light from each sub-pixel in the sensor does not correspond to the amount of red, green and blue in the resulting image. Due to the varying qualities of light and the hit and miss nature of colour temperature setting for most light sources, I suspect that images just as often have a slight magenta cast as green, though when it is a green cast it looks much less visually pleasing. mmj (talk) 07:27, 17 February 2009 (UTC)
A clear filter is not a great "luminance" filter; green is better, and is the preferred embodiment that Bayer points out in his patent. Kodak also a lot on CMY filters; while they allow more light through, they also required bigger matrix coefficients (in the sum of squares sense) to correct the color, resulting in more noise and a net loss of dynamic range usually. I agree with Mmj that the "too green" issue has nothing to do with any of this; that's a color rendering and reproduction issue, not a sensing issue. Dicklyon (talk) 17:47, 17 February 2009 (UTC)
In my view and experience, what Marc Lacoste said can be true for some pictures. It may be up to wrong demosaicing, but Mr Bryce Bayer did not do much to clear this matter either. Implementing two green sensors, instead of one, is likely driven by the constrains of a need (?) for a rectangular matrix, or better, rectangular *pixels*. But in the end, all said in regard of how luminosity and chrominance work and how human beings are mostly sensitive in green is non-sense, being mostly excuses; we, unfortunatelly, have a faulty design being all over the market, with *spatial* resolution flawed in favor of green. Similarly, using a sole green pixel, out of two, for enhancing the luminosity of the total value of the remaining rgb pixels is illogical. All in all, I suggest going with a Foveon sensor for photography (not the newer Quad) or with a 3 chip solution for camcorders.

I reverted this edit about color blindness and tetrachromacy: (Except for the 10% who are colorblind. Some humans have 4 color receptors, as well, and thus having filters such as RG1G2B also makes sense (see below), since four positions are available in each 2x2 section.) If there's some true to the idea that the eye's "greater resolving power with green light" is not so in colorblind people, that would be interesting to know and to report (with a source). As to using two different greens, that can be useful for better color accuracy, but there's not evidence that it could have any useful relationship to hypothetical tetrachromacy, espcially since all the images are going to be rendered to a standard RGB space anyway; so unless some source talks about this, we shouldn't either. Dicklyon (talk) 18:11, 13 November 2008 (UTC)

Pixel count[edit]

If I have a so-called 1MP color camera, does it have ~1M photosites (0.5M green, 0.25M each red and blue) or does it have 4M photosites? —Ben FrantzDale (talk) 23:32, 12 July 2010 (UTC)

Sorry, it doesn't have 4M of anything; just 1M total of R + G + B pixels. Dicklyon (talk) 02:38, 13 July 2010 (UTC)
Thanks. That's what I figured. While I'm not surprised, it strikes me as deceitful to say that 1e6 photosites => 1e6 pixels since that will display on 3e6 display elements, and since the Nyquist limit for green is that of an 0.5e6-pixel sensor, and is worse for B and R. Either way, thanks. —Ben FrantzDale (talk) 12:28, 13 July 2010 (UTC)


How do sensor manufacturers create millions of precisely-placed color filters aligned to sub-micron precision with the silicon? I can only imagine it's a lithographic process, but wow! Also, what kind of filter is it? Is it a dichroic filter or a conventional color filter? —Ben FrantzDale (talk) 13:59, 4 March 2011 (UTC)

Answer: Color_filter_array#Manufacture_of_the_CFA. —Ben FrantzDale (talk) 14:20, 4 March 2011 (UTC)