Encoding information with colour alone fails a meaningful slice of your audience. Roughly 1 in 12 men and 1 in 200 women have some form of colour-vision deficiency (CVD). Two hues that look obviously different to you can merge into the same tone for them — and a legend, status badge, or line chart built on those hues becomes unreadable. Paste your palette and this tool simulates all 8 CVD types to surface exactly which pairs collapse together.
How it works
For each colour the tool performs a physiologically-grounded simulation:
- The sRGB colour is linearised and converted to the LMS cone-response space.
- A dichromat projection matrix for the affected cone (long / medium / short) removes that cone’s contribution, modelling protanopia, deuteranopia or tritanopia. The result is mapped back to sRGB.
- For the anomalous (partial) variants the simulated result is blended with the original by a severity factor, modelling reduced rather than absent cone function.
- Achromatopsia is modelled by collapsing the colour toward its luminance.
Each pair of simulated colours is then compared with a CIE76 ΔE in Lab space. When that distance drops below roughly 12, the two colours are treated as indistinguishable for that deficiency and the pair is flagged.
Example
A classic data-viz mistake is pairing a #e6194b red with a #3cb44b green for “bad” and “good”. Under deuteranopia both simulate to a similar olive-brown, so a deuteranopic viewer sees one undifferentiated colour. The checker flags that pair and you can re-map green to a blue, or add a shape/label cue.
Tips
- Aim for palettes that vary in lightness as well as hue — luminance differences survive every CVD type, including total colour blindness.
- Blue-orange and blue-red pairings are usually safer than red-green.
- Even a “safe” palette should pair colour with a second cue (icon, pattern, direct label). Colour is a reinforcement, never the only channel.
- Everything is computed locally — paste confidential brand colours freely.