Color Blindness Simulator

What color blindness actually changes

Human color vision runs on three cone types tuned to long, medium and short wavelengths. When one type is missing or shifted, the brain still sees — but an entire axis of color difference disappears. In the common red–green deficiencies (protanopia, deuteranopia), reds, greens, oranges and browns fold onto one another; in the rare tritanopia, blues and yellows merge. Roughly one in twelve men and one in two hundred women carry some form — in any audience of decent size, someone sees your design differently.

The model behind this simulator

The simulation applies the transformation matrices published by Machado, Oliveira and Fernandes (2009) — a physiologically-based model derived from the stage theory of human vision, and the de-facto standard in scientific visualization tools. The matrices are applied at full severity in linear RGB, which is what the ΔE column quantifies: how far each color travels under the deficiency. A small ΔE means the color barely changes — it was never carrying red–green (or blue–yellow) information to begin with.

The palette check is the important one

A single simulated color is trivia; a pair is a decision. Interfaces and charts fail not when a color changes appearance, but when two colors that encode different things land on the same appearance. The palette mode takes every color you have saved in the dock, simulates all pairs, and flags the dangerous ones: clearly distinct for typical vision, nearly identical under a deficiency. Fix flagged pairs by separating them in lightness — lightness contrast survives every form of color blindness — or by adding a second channel entirely: labels, icons, patterns, position.

Practical rules that hold up

• Never encode meaning in hue alone; the classic red/green “bad/good” pair is invisible to the largest CVD group.
• Check your work in grayscale first — if it survives that, it survives most deficiencies.
• Prefer color ramps that vary lightness monotonically (dark-to-light) for data visualization.
• Test the final rendered artifact in image mode here — screenshots catch what palette theory misses.

Everything — including the image simulation — runs in your browser; photos and screenshots are never uploaded. When a flagged pair needs fixing, the shades generator gives you lighter and darker variants that keep the hue, and the contrast checker verifies text remains readable.