I can copy the red, green, and blue channels in an image and make them layers.

What I'm after is to have a document with 3 layers... one representing the red channel, one for green, another for blue. And when I choose the right blending mode for these layers, I end up simulating the original image... the colors will look correct.

Another possibility occurred to me, and that's to sort of simulate a bayer filter. like creating a photoshop pattern of 4 pixels. The red layer's pattern would have a white pixel at the top left, the other 3 dark. The blue layer's pattern would have the white pixel at the bottom right. The green layer pattern would have both the top right and bottom left pixels white, the other 2 dark.

How can I essentially combine pure red, green, and blue info in 3 layers to create full color?

  • Can you elaborate on why you are trying to do this? – mattdm Jul 4 '16 at 18:04
  • It's just for fun. – CreeDorofl Jul 4 '16 at 18:16
  • 1
    Let me rephrase :) — do you want to actually simulate the demosaicing, or are you mostly interested in color blending? These are very different. – mattdm Jul 4 '16 at 18:19
  • I would be interested in either of two simulations using layers: 1. 3 "plates" similar to what PS shows in the channels palette, which can be layered to make a full color image. Like the upper half of this image... showing the red channel only, and then the red+blue: i.imgur.com/c9v5mGc.jpg 2. Simulating an uninterpolated RAW using 4-pixel patterns. Something like the lower half of the previous image. I can sort of imagine simulating this in photoshop, but when I try I don't get real color, no matter what blend modes I try. – CreeDorofl Jul 4 '16 at 18:37
  • If you do this for fun and your own education, take a look at Rawdigger. It will show you lots of information about your raw files - not in simulation, but the real thing. – MirekE Jul 4 '16 at 18:50

When talking about blending the layers, creating 3 layers with red/green/blue data and choosing a blend mode of Difference can achieve what you want.

Blending RGB

  • this looks close to what I'm after, except if I create layers by copy/pasting R G and B channels from a photograph, tinting them pure red/green/blue, and then changing the blend mode to difference... I don't get these results. i.imgur.com/ikBqxDt.jpg – CreeDorofl Jul 4 '16 at 18:50
  • Just tried it on a real image and it reproduced the image exactly. I used Multiply to apply R/G/B on the b&w channel layers. – D. Jurcau Jul 4 '16 at 18:59
  • @CreeDorofl The reason you can't get those types of results is because your three source images obtained using PS channels palette aren't the equivalent of the respective values from R, G, and B filtered pixel wells. They're post-demosaicing results that already have gamma correction applied to the linear values of the sensor readout. When you boost or lower one color channel more than the other you are effectively applying different gamma correction values (and any other light curve modifications) to each color channel independently. – Michael C Jul 5 '16 at 7:25
  • @MichaelClark delayed reply but that makes sense... is it possible to access the pixels before mosaicing? Maybe Rawdigger or some other app? – CreeDorofl Oct 21 '16 at 14:10
  • dcraw allows you to view the pixels before demoisaicing. – D. Jurcau Oct 21 '16 at 14:19

Early color photography exposed three black & white plates (film) one to red, one to green, one to blue. The exposures were made via Wratten 47B blue – Wratten 58 green – Wratten 25 red. These are the three “Tri-Color” filters used to make color separations. We ended up with three negatives that represented the three primary colors. These were printed via an enlarger onto film or paper. These are “narrow cut filters. The filters overlaid on the photosites must be these or their cousins: Wratten 98 blue, Wratten 99 green, Wratten 25 red. These were used on the enlarger to make separation exposures.

The Wratten catalog numbers come from the master filter maker Frederick Wratten (England firm of Wratten and Wainwright, purchased by Kodak 1912. The Wratten cat # retained. Today these filters are obtainable, likely the name Wratten has been dropped.

Any way, you can photograph three images via these filter sets. The Bayer matrix likely uses these same filter values. On the Bayer matrix, the proportion of red to green to blue filtered photosites is based on the sensitivity of the chip. You should be able to accomplish your task by making three images using these filters values.

  • I wasn't really looking to spend money on this and take additional photos. I'd just like to simulate RGB channels as layers or simulate an uninterpolated RAW image. Something like this -> guillermoluijk.com/article/virtualraw/bayer.gif Except of course a raw has 2 greens for 1 R and G. So it looks like a greenish cast. – CreeDorofl Jul 4 '16 at 18:19
  • A true representation of that example would be in various shades of gray with no color. The different color filters alter the luminance values based on the color passing through them, but they're all still only brightness values. Just because light registers to a green filtered pixel doesn't mean all of the light getting down the pixel well is green. Some yellow and aqua will also pass through the green filter, though at lower intensities than a green light from an equally bright source. The filters in front of the pixel wells are just like color filters in front of B&W film. You still get B&W – Michael C Oct 22 '16 at 7:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.