How to use the Color Blindness Simulator
- Drop an image on the upload box or click it to choose one — a chart, UI screenshot, map or photo. Or keep the built-in sample palette.
- Choose a type of colour vision deficiency. For the anomalous types (-anomaly), set the severity.
- Drag across the image or use the split slider to compare the original (left) with the simulation (right).
- Scan the All eight types grid to spot any colours that merge, and click a thumbnail to inspect it large.
- Press Download to save any simulation as a PNG for your design review or bug report.
How this colour blindness simulator works
Most people have three types of cone cells, sensitive to long (L, “red”), medium (M, “green”) and short (S, “blue”) wavelengths. Colour vision deficiency (CVD) happens when one cone type is missing (dichromacy, the “-opia” types) or its sensitivity is shifted (anomalous trichromacy, the “-anomaly” types). This tool recalculates every pixel of your image to approximate what a person with each condition would see.
Method and formulas
The red–green and blue–yellow simulations use the published matrices of Machado, Oliveira and Fernandes (2009), “A Physiologically-based Model for Simulation of Color Vision Deficiency” (IEEE Transactions on Visualization and Computer Graphics). The paper tabulates a 3 × 3 matrix for each type at severities from 0.1 to 1.0; severity 1.0 models dichromacy. For in-between slider values the tool interpolates linearly between the neighbouring published matrices.
Because these models describe light, not screen values, the matrices are applied in linear RGB:
2. Simulate: [R′ G′ B′] = Mtype, severity × [R G B]lin
3. Re-encode: C = 12.92·Clin if Clin ≤ 0.0031308, else 1.055·Clin1/2.4 − 0.055 (clamped to 0–1)
Achromatopsia (complete colour blindness) is simulated by replacing each pixel with its relative luminance, Y = 0.2126 R + 0.7152 G + 0.0722 B in linear RGB. There is no widely agreed physiological model for achromatomaly (partial colour blindness), so it is approximated by blending each pixel toward that luminance by the severity you choose. Every row of the Machado matrices adds up to 1, so white stays white, black stays black and neutral greys are unchanged.
Worked example
Pure red, sRGB (255, 0, 0), is linear (1, 0, 0). The protanopia matrix maps it to linear (0.152, 0.115, −0.004), which re-encodes to roughly (109, 95, 0) — a dark olive-brown. Its luminance falls from 0.213 to about 0.114, which is why bright red text or warning icons can look dark and dull to people with protanopia.
The eight types at a glance
| Type | Cone affected | What gets confused |
|---|---|---|
| Protanopia / protanomaly | L (red) missing / weak | Red with green, brown and orange; reds look darker |
| Deuteranopia / deuteranomaly | M (green) missing / weak | Red with green, brown and orange; the most common type |
| Tritanopia / tritanomaly | S (blue) missing / weak | Blue with green, yellow with pink or grey; rare |
| Achromatopsia | No working cones (rod monochromacy) | All hues; only lightness is seen |
| Achromatomaly | Reduced colour perception | Colours look washed out |
Red–green deficiencies are by far the most common, affecting roughly 8% of men and 0.5% of women of Northern European ancestry, with lower rates in many other populations. Tritan defects and achromatopsia are rare.
Designing for colour-blind users
- Do not rely on colour alone. Pair colour with text, icons, patterns or position — WCAG 2.x success criterion 1.4.1 (Use of Color). A red/green “error/OK” dot should also have a label or symbol.
- Check contrast. WCAG asks for at least 4.5:1 for normal text, 3:1 for large text and 3:1 for interface components and meaningful graphics. Good lightness contrast survives every type of CVD.
- Use colour-blind-safe palettes for charts, such as Okabe–Ito or the Viridis family, which vary in lightness as well as hue.
- Label lines and bars directly instead of using a colour-only legend, and use different dash styles or markers.
- Underline links in body text rather than only colouring them.
- Test real screens. Run chart exports, dashboards, maps and status indicators through the simulator before release.
Limitations
No simulation can show exactly what a specific person sees: colour vision varies between individuals, and the result also depends on your monitor’s calibration. Simulations are a design aid for spotting problem colour pairs, not a diagnosis. If you think you may have a colour vision deficiency, see an optometrist. Large images are processed at up to 2,400 pixels on the longest side to keep the page fast; the original file is never uploaded. To grab exact colours from your design, use the colour picker, and resize screenshots first with the image resizer.
Frequently asked questions
How accurate is this colour blindness simulator?
What is the difference between protanopia and protanomaly?
What is the most common type of colour blindness?
Are my images uploaded?
How do I make a chart readable for colour-blind people?
Why does white stay white in the simulation?
Last reviewed November 2026. Results are estimates for information only — see our disclaimer.