RYB coordinates are barycentric — normalise them and every pigment lands on a triangle with red, yellow and blue at the corners. Distance from the centre is chroma, and straight through the centre is the complement. That is why mixing a colour with the thing opposite it makes mud: you are walking through the middle.
The tool answers “what does this make?”. This asks the other way round: give it a colour and it searches every two- and three-pigment mix for the closest it can get.
#4a7b9d, 74, 123, 157
or 210 40% 45% and the rest follow. CMYK here is the naive conversion,
not a press profile: real four-colour printing depends on ink, paper and press, and no
formula can tell you what a given printer will produce. Lab is read-only because it
describes a colour rather than specifies one.
Each pigment absorbs some wavelengths and reflects the rest, so mixing removes more light than either alone. Screens add light; paint does the opposite.
The model artists are taught, built from eight reference points (white, black, three primaries and their pairwise mixes) and blended between them.
Two adjacent primaries make a secondary (orange, green, violet); a primary with its neighbouring secondary makes a tertiary.
Real pigments have their own reflectance curves (Kubelka–Munk theory), opacity and tinting strength. Test on a real tube.
Colours opposite each other on the wheel (red and green, blue and orange, yellow and violet) cancel out the most when mixed. Useful for muting a colour deliberately, a trap if you wanted something vibrant. Every real pigment also absorbs some light outside its main colour, so mixtures trend toward brown or grey the more colours you add.
Some pigments (phthalo blue) are so strong a tiny amount overwhelms a mix; others (many earth tones) are gentler. This tool mixes by simple parts, a reasonable starting approximation, not a substitute for testing a real tube.
A tint (adding white) raises opacity and lightness together. A wash (thinning with medium) lightens the visual effect without changing the underlying mix the same way: different technique, different result. A shade is a colour with black (darker, often muddier), a tone one with grey.
A target colour from two base paints is a ratio problem: the same two paints in a different ratio give a completely different, unrepeatable shade, and without measuring, matching it again for a touch-up becomes nearly impossible.
Every real pigment absorbs some light outside its "main" colour too — mixing pigments compounds those absorptions, which is why paint mixtures trend toward brown or grey the more colours you add, and why a "pure" digital blue looks more muted once it's modelled as a real pigment.
No — real pigments have their own reflectance curves (the basis of the more rigorous Kubelka–Munk theory used in industrial colour matching), and opacity, tinting strength and paint chemistry all affect the result. This is a reference approximation of subtractive mixing behaviour, not a colour-matching instrument.
A tint is a colour mixed with white (lighter, less saturated). A shade is a colour mixed with black (darker, often muddier since black pigment absorbs broadly). Both are separate from a tone, which mixes in grey.
Mixing colours from opposite sides of the colour wheel (like red and green) pulls the result toward brown or grey, since their absorptions overlap the least useful way — this is the same reason painters are taught to avoid complementary-colour mixes when they want a clean, vibrant result.
Mixing by simple parts, and equal tinting strength between the colours — which real pigments emphatically do not have. A strong phthalo blue will dominate a weak yellow far beyond its share of the mix, and opacity varies just as much, so a transparent glaze behaves differently again from an opaque one. Predicting that properly needs measured reflectance data for the specific paints. Screen colour is also emitted light rather than reflected pigment, so treat the swatch as a direction of travel and test on scrap before committing.
The gamut readout and the hue chart measure this model, not a paint range. That matters most at the warm end: the model places its red and yellow primaries exactly on your screen’s red and yellow, so a saturated orange reports as fully reachable by construction rather than by measurement — real pigment sits inside that, and a tin sits inside again once it dries. The cool end is the honest half, where mixing genuinely runs out of chroma. Differences are quoted as CIE76 ΔE, which is a plain distance in Lab and overstates the gap in saturated blues, so read the ΔE as a ranking rather than a precise perceptual score. CMYK here is the naive conversion, not a press profile. The hatched area on the picker is drawn from the same model: it is the boundary of what this palette can mix, not a statement about paint in general — a manufacturer with a pigment we do not carry will reach past it. It is drawn as two lines rather than one because the edge is not a cliff: a colour just outside is very nearly mixable, and the gap between the lines is the width of that doubt rather than a measured tolerance.
Releases in which this page changed, newest last. Derived from the archived copy of every release, not from notes written afterwards — so it reflects what actually shipped. Site-wide passes are left out; they are in the full changelog.