Colour Blindness Simulator
See a palette as people with protanopia, deuteranopia, tritanopia or achromatopsia see it (Machado 2009 matrices), and find the pairs that become confusable.
Every colour beside its four simulated versions, the closest pair under each condition with an OKLab distance before and after, and a flag when two colours collapse into one.
Example: Pure red #ff0000 becomes #6d5f00 under protanopia and #a39000 under deuteranopia; against #808080 it is indistinguishable in greyscale (ΔE 0.3), so a red/grey status colour scheme fails.
A published model,
applied honestly.
What the matrices are, how the colours are transformed, and what the result does and does not say.
The model
Machado, Oliveira and Fernandes (2009) derived 3 × 3 matrices that transform linear RGB into what a viewer with a given cone deficiency sees, from a physiological model of the cones. The page uses their full-severity matrices for protanopia (no L cones), deuteranopia (no M cones) and tritanopia (no S cones); achromatopsia is shown as luminance.
The transform
Each colour is linearised (sRGB transfer function), multiplied by the matrix, clipped and re-encoded. Distances between colours are measured in OKLab, so "closest pair" reflects perceived difference; a pair that was well apart but ends within ΔE 5 is flagged as confusable.
What it means
The matrices model complete dichromacy; most colour-deficient people have milder anomalous vision that sits between the original and the simulation, and screens and lighting vary. Use the result to find risky pairs and to add a second cue (shape, text, pattern), not to certify a palette. Nothing leaves the browser; the same four anonymous usage counts as the rest of the site apply.
SOURCES
- Machado, Oliveira & Fernandes, "A Physiologically-based Model for Simulation of Color Vision Deficiency", IEEE TVCG 15(6), 2009 — severity 1.0 matrices applied in linear sRGB
- sRGB transfer function (IEC 61966-2-1) and Rec. ITU-R BT.709 luminance coefficients for the achromatopsia view; distances in OKLab
Last reviewed 20 September 2026. How results are checked: How we verify.