Fellow photographers and I were discussing how to create a positive colour image by overlaying three film BW negatives/positives that were shot using the trichrome process, using three colour filters in between the stacked photos.

We concluded that to make this work, one has to do three things:

  1. Use RGB filters when shooting,
  2. use BW negative film,
  3. use CMY filters in the stacked final product, respectively

This is a case of subtractive colour mixing, which is based on cyan, magenta, and yellow, which combined at 100% 'strength' filter out all light.

Looking at the colour wheel below, one can see that CMY and RGB both are opposing colours, all equally far apart. This has me wondering: why is CMY specifically used for subtractive colour mixing? Won't red, green, and blue not result in black just as much, since they are in equal relative positions on the colour wheel? And what about other colours, such as pink, light blue, and light green? In fact, would CMY work in an additive setting?

enter image description here

  • @MichaelC did you really have to change every instance of 'colour' to 'color'?
    – timvrhn
    Aug 17, 2021 at 7:06
  • @MichaelC I understand from what you have linked that both RGB and CMY stem from the RGB cones the human eye has. After all, CMY is based on RGB too in that cyan absorbs red but emits green and blue, etc. Would it be sensible to conclude that in theory, for both additive and subtractive colour mixing, other sets of colours could be used for eyes with different cones of other sensitivities?
    – timvrhn
    Aug 17, 2021 at 7:16
  • 1
    The cones in the human retina are what you term "cones of other sensitivities!" They are not really most responsive R, G, & B, respectively. They are most responsive to 'blue-violet', 'green with just a touch of yellow', and 'yellow-green' that are not evenly spaced around the color wheel. The peak response of M-cones ("green") and L-cones ("red") are only 30 nanometers apart at ≈534nm and ≈564nm, respectively. S-cones are most responsive to ≈420nm. 'Red' is 640nm, 'Green' is 525-530nm, and 'Blue' is 480nm.
    – Michael C
    Aug 17, 2021 at 20:43
  • 1
    Re human retinal cone sensitivities: en.wikipedia.org/wiki/CIE_1931_color_space#/media/…
    – Michael C
    Aug 17, 2021 at 20:50

3 Answers 3


You have your causality backwards. RGB doesn't work because they're opposing colors on the wheel, the wheel is organized as it is because the eyes work in RGB.

The way we see colors is an additive process. You can create (almost1) any hue by adding different quantities of red, green, and blue light. There are no other colors of light that have that property.

The reason CMY is used in printing is that those colors absorb one of the 3 primaries while reflecting the other 2. Cyan absorbs Red, Magenta absorbs Green, and Yellow absorbs Blue. To print something that's red, you combine magenta and yellow so that all the green and blue light hitting it gets absorbed, leaving only the red.

1 You can't get all the hues because the response of the eye as defined by the CIE color space is not perfectly triangular; some hues can only be seen with pure monochromatic light at specific frequencies. RGB gets close enough for practical use though.

  • Thank you Mark, you captured what the other answers were hinting at. You hit my misconception on the head.
    – timvrhn
    Aug 18, 2021 at 8:17
  • More questions – Few answers Dr. Edwin Land, working on Poloid’s color film project, repeated Maxwell's three color projection method. When dismantling the apparatus, the blue projector was turned off first. To everyone’s amazement, the audience continued to see a full color image. Thus, he published a two color theory. He never was able to apply this approach to film photography. Aug 18, 2021 at 15:16

At the time of taking the images you would want to use a combination of RGB filters; e.g. R+G rejection filters (or a blue colored absorptive type filter) allowing B to pass.

Then at the time of enlargement/printing you would need to use the corresponding RGB filter for that image; e.g. B for the R+G filtered negative... this is all additive color because it all has to do with projecting/recording light.

And the color paper has RGB sensitive layers which are CMY in color correspondingly (i.e. the cyan layer is sensitive to red light)... The final output is in subtractive color because the positive reflects/transmits the colors it does not absorb/subtract.

The other colors (i.e. magenta) would work at the additive stage because it is essentially a combination of the R+B positive colors (typically labeled as violet in the visible light spectrum).

At any stage involving projecting light you are using additive color (RGB). Additive colors add up to white, so a white (or colored) light falling on a reflective surface (white wall) determines the color.

Subtractive colors (CMY) have to do with reflecting light, and they add up to black. What makes something green is that it absorbs all of the other colors. I.e. it is a mix of cyan and yellow (subtracting red/blue) and reflects the remaining additive color in the light (green). I.e. you cannot mix the processes... light is always additive, and surfaces are always subtractive.

  • Steven, in my question I refer to stacking three film frames in combination with colour filters, not printing. Furthermore, I believe printing should be done in the same manner as what I describe in my post. That is, shooting with RGB filters and printing using CMY filters for the respective frames. As you're dealing with a double negative, printing a red channel using a cyan filter will result in red colours on the print, etc. Besides that, this doesn't address or answer my question.
    – timvrhn
    Aug 16, 2021 at 17:41
  • Steven, upon rereading your question I see why your method would work. Do you have a suggestion why my proposed method doesn't work, or would both work equally fine?
    – timvrhn
    Aug 16, 2021 at 17:51
  • 1
    @timvrhn, I've added to my answer to help clarify. Your cyan filter for printing (light projection) is actually a combination of green and blue absorption filters. And you have to be careful as to the type of filter... e.g. a red absorption filter appears red because it absorbs all of the other colors (subtractive color); whereas a red rejective filter appears red on the rejection side, and cyan on the transmission side. Aug 16, 2021 at 20:16
  • 1
    Keep in mind that the traditional color wheel isn't an accurate representation of human color sensitivity. Use of a CIE color space model (and understanding where spectral colors do and, more importantly, do not fall in CIE color space) can be enlightening in this regard. Also please keep in mind that color only exists in the perception of the eye-brain system that "sees" it. There is no specific color implicit in any wavelength of elecromagnetic radiation, in out of the "visible spectrum".
    – Michael C
    Aug 17, 2021 at 20:53
  • 1
    Other species' vision systems perceive wavelengths humans do not, and some do not perceive wavelengths humans do. Dogs, for instance, do not perceive the same colors that humans do for the wavelengths our vision systems have in common.
    – Michael C
    Aug 17, 2021 at 20:57

The world saw the first color image by photography in 1861 when James Clark Maxwell demonstrated his three-color additive method to the Royal Society. Gavriel Lippmann got the Nobel Prize for his no-filter no-dye color interference process 1908. The first commercial color film, Autochrome was marketed in 1903, using red, green, blue dyed microscopic flakes of potato starch invented by Auguste and Louis Lumiere.

Since those successes, numerous methods using additive (red, green, blue) and subtractive (cyan, magenta, yellow) have been marketed.

Maxell took three successive pictures, a still life, each taken with one of the three additive color filters. The film, developed as a positive and each simultaneously projected using three projectors each filtered with one of the three additive color filters.

Lippmann made a super transparent film emulsion. This glass plate emulsion was backed by a mirror. The exposing light traversed the film emulsion exposing the plate. The exposing light hit the mirror and again exposed the film from the rear. When this processed plate was illuminated, light traversed and found its way to an observer’s eye. This light was forced to filter through a maze of black and white silver (images). Only those frequencies that originally exposed the film could pass. A full color picture resulted.

Modern color films are exposed via filters or by adjusting what frequencies of light they are sensitive to. With few exceptions film is sensitive to three light additive colors of red, green, and blue. During developing, the three silver black and white images are bleached away. The film layer exposed via red light is replaced by cyan dye. The Blue sensitive black and white image is replaced with yellow dye. The green sensitive black and white image is replaced with magenta dye. The three subtractive primary colors are the complements (opposite) of the three additive primary colors.

The idea is to present to the viewer a faithful reproduction by controlling the intensity of the three additive primaries the observer will see. To accomplish we need dye or pigment that will filter (control) how much red, green, and blue light is seen.

A cyan filter blocks red light. Thus, this dye in the film controls how much red light will traverse. A yellow filter blocks blue light. Thus, a yellow filter dye in the film controls how much blue light will traverse. A magenta filter blocks green light. Thus, this dye in the film controls how much green light will traverse.

The key here is that cyan filter blocks red and transmits green and blue. A magenta filter blocks green and transmits red and blue. A yellow filter blocks blue and transmits red and green. The secret is, the subtractive primaries filter out one of the primaries and pass two primaries. This is the simplest of methods. This method works for slides (transparencies), color negative films, and for prints on paper.

The biggest problem is finding dye or pigments that have the “right” color (not easy). The yellow dye is excellent, the magenta is fair, the cyan dye is poor. For prints on paper, the three should overlap to form black. The black we get is not deep enough, we must add a fourth black dye called a “kicker” to get the a good black. For photographic film and photographic paper, we never got this right; we are forced to live with a poor black.

By the way, if you superimpose the three negatives or three positives, the three silver images will be super dense i.e. opaque. You must bleach away the images and then substitute transparent dye. Plus - Stacked CMY = black i.e. no light will make it through the sandwich. The cyan filter blocks red, the magenta filter blocks green, the yellow filter blocks blue. Stacking CMY blocks all three of the light primary colors. This will not work!

  • Especially thanks for the last paragraph Alan, you are absolutely right. Initially I envisioned the dense parts of the negatives would block out a specific colour (be it cyan, magenta, or yellow), thereby ensuring the end product isn't 100% of the three colours (and thus, black). However, this density itself of course also diminishes light intensity.
    – timvrhn
    Aug 18, 2021 at 8:12

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.