On film, a colored filter sits in front of the lens and changes which wavelengths reach the emulsion before exposure. A digital sensor records full color first, so the equivalent control happens after capture: the three color channels are weighted and summed into a single gray value. A channel mixer performs exactly this sum, which is why it can imitate a yellow, orange, or red filter without any glass. The imitation is close but not identical, and the difference is set by how the sensor saw color in the first place.

How a Channel Mixer Reproduces Filtration

Every grayscale conversion reduces three numbers to one. The default weights are not arbitrary. The luma coefficients in ITU-R Recommendation BT.709 assign 0.2126 to red, 0.7152 to green, and 0.0722 to blue; the older BT.601 standard used 0.299, 0.587, and 0.114. In each case the weights sum to one, which preserves overall brightness and matches the eye’s strong sensitivity to green and weak sensitivity to blue.

A physical filter passes its own color and absorbs the complementary one, lightening like tones and darkening opposite ones. A red filter on film darkens blue sky and renders red objects light. The channel mixer reproduces this by raising the red weight and lowering the blue: a pixel of blue sky carries a high value in the blue channel and a low one in the red, so emphasizing red while suppressing blue drives that pixel dark. Weighting the red channel alone is therefore the software analogue of a deep red filter, and a mix biased toward green and red, with blue reduced, approximates the gentler effect of a yellow filter.

Why the Match Is Imperfect

The limit on this emulation is the color filter array over the sensor. In a Bayer pattern, each quad of photosites carries one red, one blue, and two green filters, and demosaicing interpolates the missing values. These dye filters are broad and overlapping rather than sharp cutoffs, so the red channel responds partly to green and even infrared wavelengths, a behavior described as spectral crosstalk in the image-sensor literature. A glass red filter on a panchromatic film blocks blue light at the source; the red channel, by contrast, has already absorbed a blend of wavelengths that cannot be separated afterward. Information a physical filter would have excluded before exposure is already mixed into the recorded values, so the conversion redistributes existing data rather than removing light. Once a region clips to white or buries detail in shadow, no weighting recovers it.

Practical Consequences

Because the weights act on data already present, channel mixing is reversible and free of the exposure penalty filters impose. A deep red filter on film costs roughly three stops; the mixer costs nothing. Two constraints follow from preserving tonality. Keeping the weights summed near 100 percent holds midtone brightness steady, and extreme settings amplify channel noise, since the blue channel in particular is often the noisiest. Color cannot be invented after the fact: where a scene lacks separation between, say, a red flower and green foliage of similar luminance, the mixer can exaggerate the difference far more aggressively than glass, but only to the extent the sensor distinguished the two colors at capture.