What to see
To some degree, all colours could be called illusions, at least they are a fascinating visual phenomenon. Here you can view and experiment with additive colour mixture (left) and subtractive colour mixture (right).
What to do: mixing
The “autorun” feature produces seemingly random effects. To explore, start by setting to “white” with a preset selector. View the left RGB colour sources as theatre spotlights, which at full energy mix to white. Varying the individual sources produces all possible hues and saturations. E.g. reducing the blue spotlight to zero leads to a saturated yellow by mixing additively red and green.
Less intuitive for most will be "subtractive” colour mixing. Imagine it as follows: The cyan filter is either transparent (slider at bottom) or fully in the beam, letting pass only green and blue, rejecting red. A magenta lets only red and blue through, rejecting green. The yellow filter passes only red and green, rejecting blue. Thus no light will come out when all three filters are fully in. This is, somewhat misleadingly, calle “subtractive colour mixing”.
Finally, you may want to try to first mix a colour additively and then send it through the filters. That is what happens when you look at a photo printout from your inkjet printer, which uses subtractive mixing under tinted room illuminaction.
Colour vision is complex and a wide field where many aspects interact, especially when it comes to art. One inescapable “bottleneck” for all colour perception is our eye, where all wavelength combinations (plus neighbor effects) ultimately end up as an activity combination of the three photoreceptor types: red (should be called reddish yellow, L for long wavelengths), green (actually a greenish yellow, M for “middle”) and blue (only ~5% of all receptors, S for short wavelengths). Thus colour vision is 3-dimensional, and can be additively described by an RGB combination, subtractively by Cyan/Magenta/Yellow or more phenomenologically by luminance, saturation, hue.
Additive colour mixing happens on your typical computer monitor. If you look at the screen with a magnifying glass, you will discern the single RBG colour contributions, and with larger distance they blend into each other.
Subtractive colour mixing is used in printing, where the layers of pigment above each other resemble our filters here. Usually, to get deep black, black is added as a forth colour since real filters are not as steep as the ones simulated here. By the way: Subtractive colour mixing is a historical misnomer, it should be called “multiplicative” mixing, since per wavelength the incoming intensity is multiplied by the filter transmission at that wavelength.
Finally, if you are an experienced artist, your colour mixing experiences will differ from both additive and subtractive. There are two reasons:
(1) paint pigment mixing is a mixture of additive and subtractive: the surface mixes additively, light passing through the pigments, bouncing off the ground suffers subtractive mixing. [Fig. 6.2 by David Briggs.] This is sometimes called “partitive mixing”, but the nomenclature is not very consistent. Colour mixing in painting is a very complex field and not the topic of the present demonstration.
(2) Different pigments may look like having nearly the same colour, but actually transmit quite different spectra. This is due to the fact that we all are, to a degree, colour blind: widely different spectra look alike (they are called “metamers”), as long as they evoke the same activation triplet in the 3 receptor types.
Wyszecki G, Stiles WS (1982) Color science: concepts and methods, quantitative data and formulae. Wiley, New York
Goldstein BE (2009) Sensation and Perception. Wadsworth Publishing Company, ISBN-13: 9780495601494