How a color looks with protanopia, deuteranopia or tritanopia (red, green or blue color blindness), using the Machado et al. (2009) model.
By AdminDrafted with AI writing toolsUpdated Checked examples: 4
Without green-sensitive (M) cones, #E74C3C is seen roughly as #9F8F36, a shift of 0.188 in OKLab distance. Don't rely on this hue alone to carry meaning; pair it with text, shape or a lightness difference.
Machado et al. severity-1.0 matrix, applied in linear RGB as colorspacious and DaltonLens do.
Channels are clipped to 0–1 before the sRGB curve and rounded half up to 8 bits.
The simulator converts the color from sRGB to linear light, multiplies it by the Machado, Oliveira and Fernandes (2009) matrix for the chosen type of color vision deficiency, clips each channel to the 0–1 range and converts back to sRGB. The severity 1.0 matrices used here model complete loss of the red-sensitive (protanopia), green-sensitive (deuteranopia) or blue-sensitive (tritanopia) cones. The color shift is the straight-line distance between the original and simulated colors in OKLab.
Designers use it to check whether a color still carries its meaning for people with color vision deficiency. The default red, #E74C3C, becomes the olive #9F8F36 with deuteranopia, a shift of 0.188 in OKLab, and #786D39 with protanopia.
The milder anomalous forms, such as deuteranomaly, are more common than the complete ones and shift colors less, so the simulation shows the strongest effect. Displays and viewing conditions also change how colors look.
A simulator converts the color to linear RGB, multiplies it by a 3 × 3 matrix that models the missing cone type, clips the result to the 0–1 range and converts back to sRGB. With the Machado, Oliveira and Fernandes (2009) matrices, red #E74C3C becomes olive #9F8F36 for deuteranopia and #786D39 for protanopia.
Protanopia and deuteranopia both collapse the red–green axis, so reds, oranges and greens drift toward olive and khaki. Tritanopia collapses blue–yellow instead and leaves reds largely intact: #E74C3C becomes #FE1B49.
Each type uses the severity 1.0 matrix from the authors' supplementary material, which models complete loss of the cone type. The milder anomalous forms, such as deuteranomaly, are more common and shift colors less.
Order matters. Applying the matrix to the encoded 0–255 values skips the linearization and gives a different color; DaltonLens explains why the matrices act on linear RGB.
| Scenario | Simulated hex | OKLab shift |
|---|---|---|
| Default | #9F8F36 | 0.1875 |
| #E74C3C protanopia | #786D39 | 0.2047 |
| #E74C3C tritanopia | #FE1B49 | 0.0648 |
| #FF0000 deuteranopia | #A39000 | 0.2495 |
| #000000 deuteranopia | #000000 | 0.0000 |
Default is #E74C3C with deuteranopia. The shift is the straight-line distance between the original and the simulated color in OKLab, a perceptual color space.
CSS Color 4 treats an OKLab distance of 0.02 as just noticeable, so 0.1875 is a large move. Black stays black under every matrix, because zero light multiplied by any matrix is still zero.
The useful question is whether two colors with different meanings stay distinguishable. A typical error red, #E74C3C, and success green, #2ECC71, become #9F8F36 and #BAAE77 under deuteranopia: two khaki tones that differ mainly in lightness.
Pure red #FF0000 and CSS green #008000 fare the same way, becoming #A39000 and #776A18. When a pair converges like this, give it a second cue:
rgb(), hsl() and oklch().For text legibility, test the pair with the contrast checker. Simulation and contrast answer different questions, and a design needs both.
Checked against: Separate exact decimal arithmetic script: sRGB → linear (IEC 61966-2-1), Machado 2009 severity-1.0 matrix, clamp, re-encode, round half up
Checked against: Separate exact decimal arithmetic script: first matrix column (0.152286, 0.114503, −0.003882) re-encoded; blue clamps to 0
Checked against: Each Machado matrix row sums to 1 within 1e-6, so linear (1, 1, 1) maps to itself
Checked against: Separate exact decimal arithmetic script: third matrix column (−0.178779, 0.147602, 0.303900) re-encoded; red clamps to 0
About 8% of men and 0.4% of women of European ancestry have red-green color deficiency, according to a 2012 review of population surveys by J. Birch in the Journal of the Optical Society of America A; among men of Chinese and Japanese ethnicity the figure is 4% to 6.5%. The US National Eye Institute puts it at about 1 in 12 men.
Each is the complete loss of one type of cone. Protanopia (no red-sensitive cones) and deuteranopia (no green-sensitive cones) make red and green impossible to tell apart. Tritanopia (no blue-sensitive cones) confuses blue with green, purple with red and yellow with pink, according to the US National Eye Institute. The milder forms, protanomaly, deuteranomaly and tritanomaly, shift colors less.
Deuteranomaly, reduced sensitivity of the green-sensitive cones that makes some greens look redder, is the most common type, according to the US National Eye Institute. Red-green deficiencies as a group are far more common than blue-yellow ones. This simulator shows deuteranopia, the complete form of the same deficiency, which is the strongest case of that confusion.
Don't use color as the only way to convey information. WCAG 2.2 success criterion 1.4.1, Use of Color (level A), requires another visual cue such as text, a pattern or an icon, and criterion 1.4.11, Non-text Contrast (level AA), asks for a 3:1 contrast ratio between interface components or graphics and the colors next to them. Check important color pairs here under all three types.
This calculator includes 4 worked examples with independently sourced answers. They run in the test suite, and you can run them here too.