Color Blindness Simulator

Typical color vision uses three cone types: L (long wavelength, "red"), M (medium, "green") and S (short, "blue"). Color vision deficiency (CVD) happens when one type is missing (the "-opia" forms, also called dichromacy) or has its sensitivity shifted (the "-anomaly" forms, anomalous trichromacy).

Red-green deficiencies are inherited on the X chromosome, which is why they are far more common in men: about 8% of men and 0.5% of women of Northern European ancestry, with rates varying between populations. Tritan defects are not X-linked, so they affect men and women equally. Because these figures are commonly cited as rounded estimates, treat them as orders of magnitude rather than precise counts.

It uses the physiologically based model by Machado, Oliveira and Fernandes (IEEE Transactions on Visualization and Computer Graphics, 2009), which models anomalous trichromacy as a shift in a cone's spectral sensitivity and dichromacy as the limit of that shift. The authors published one 3×3 matrix per severity from 0.0 to 1.0 in steps of 0.1; this tool interpolates between them for the severity slider.

The matrices operate on linear RGB, so each pixel is first decoded from the sRGB transfer curve, multiplied, clipped and re-encoded. Skipping the linearization, a common shortcut, distorts the result. Achromatopsia maps every color to its relative luminance.

Deuteranopia (severity 1.0), linear RGB
[ 0.367322   0.860646  -0.227968 ]
[ 0.280085   0.672501   0.047413 ]
[-0.011820   0.042940   0.968881 ]

Any simulator shows a model of *average* perception. Individuals vary, and single-matrix models are least certain for tritan defects. Use it to catch design problems, never to diagnose: only an eye care professional can test color vision.

For every pair in your palette, the simulator measures perceptual distance in OKLab (scaled by 100, so it reads like the familiar ΔE). A pair is flagged when it is clearly different with typical vision but drops below your threshold after simulation.

About 2 is the smallest difference you can see between two large swatches side by side. Small marks such as chart lines, map regions and legend dots need far more, which is why the default threshold is 10. Raise it if your colors appear as thin lines or small dots; lower it for large blocks of color.

Watch the Achromatopsia row too. It shows lightness alone, so any pair it flags differs only in hue. Making those colors differ in lightness is the single most effective fix: it helps every type of color vision at once, and grayscale printing as a bonus.

The rule behind WCAG success criterion 1.4.1 Use of Color: color must never be the *only* way information is conveyed. In practice:

• Add a second cue. Pair status colors with icons and words ("Error: card declined"), give chart series different marker shapes or dash patterns, and underline links in body text.
• Label directly. Put labels on chart lines and pie slices instead of relying on a color legend alone.
• Vary lightness, not just hue. Pairs with a clear lightness difference survive every deficiency.
• Avoid problem pairs with similar lightness: red–green, green–brown, green–orange, blue–purple and pink–gray. For good and bad, blue and orange is a safer pair than green and red.
• Start from a tested palette. The Okabe–Ito palette (#E69F00, #56B4E9, #009E73, #F0E442, #0072B2, #D55E00, #CC79A7, #000000) was designed for color-blind readers; for heatmaps, perceptually uniform maps such as viridis hold up well.

Then check readability with the color contrast checker: good contrast and good color-blind design reinforce each other.