Palettes that work for every kind of color vision
Roughly one in twelve men has some form of color vision deficiency (CVD). A palette that looks fine to you might turn a green and an orange series into the same muddy color for them. This tool generates palettes and then verifies each one by simulating the three dichromacy types and rejecting any pair of colors that becomes indistinguishable.
How it works
For each candidate palette the tool runs three steps per color:
- Simulate CVD. Colors are converted to the LMS cone space and projected through the Brettel/Viénot dichromat matrices for protanopia, deuteranopia, and tritanopia.
- Convert to CIELAB. Both the simulated colors are converted to L*a*b*, a perceptually-uniform space where Euclidean distance approximates how different two colors look.
- Check every pair. It measures the CIE76 deltaE between each pair after each simulation. If the smallest distance across all pairs and all simulations stays above a safety threshold, the palette passes.
It also enforces a minimum lightness spread so colors differ in brightness too, which helps grayscale printing and the achromatic case.
Understanding the three deficiency types
Each type of dichromacy results in a different pattern of color confusion, which is why the verification must pass all three simulations:
Deuteranopia (green-cone insensitivity) is the most common form, affecting around 6% of men. The primary confusion is between reds and greens — a red line and a green line on a chart can appear nearly identical. Oranges also shift toward yellow, reducing contrast with adjacent colors.
Protanopia (red-cone insensitivity) is less common (around 2% of men) but more severe in its effect on luminosity. Red appears very dark — almost black — in addition to being confused with green. This means high-saturation red used as an “alert” color may disappear entirely for someone with protanopia.
Tritanopia (blue-cone insensitivity) is rare and affects blue-yellow discrimination. Blue and green become similar, and yellow and pink become difficult to tell apart. The confusion pattern is quite different from the red-green types, which is why palette checking must include it separately.
What the palette actually looks like
Palettes that pass CVD verification tend to rely on:
- Blues and oranges — well-separated across all three deficiency types
- Lightness variation — different luminosities make colors distinguishable even in grayscale
- Avoiding red-green pairs — the most common failure mode
You will rarely see two saturated reds and greens in the same verified palette. Instead expect blues, oranges, purples, and varied-lightness neutrals. This is not a limitation but the correct result — those are the combinations that actually remain distinguishable.
Beyond the palette: accessibility in practice
The palette is a starting point, not a complete solution. Effective accessible visualization uses color as one of several simultaneous cues:
- Labels directly on lines or bars (not only in a legend) eliminate the need to decode color at all
- Distinct shapes for data markers (circle, square, triangle) per series
- Pattern fills in bar charts (solid, hatched, dotted) that print distinctly in grayscale
- Direct annotation of key data points in the text or caption
A chart readable by someone with no color perception at all is universally more readable — it works on grayscale printouts, on small screens, and in projected environments where colors shift.
Tips and notes
- Always add a second cue. Labels, dashes, and markers make a chart readable even for someone who turns off color entirely.
- The simulation is the standard dichromat model, so it represents the severe end of each deficiency — passing here means anomalous (milder) cases are safe too.
- Smaller palettes are easier to keep safe. If you need many series, consider grouping or faceting instead of cramming colors together.