How to use it
- Click Choose a file and pick a picture, or drop it on the box. The original appears on the left.
- Choose the Deficiency: protanopia or deuteranopia.
- Click Simulate and download. The simulated picture appears on the right and the JPEG downloads.
The model and its two matrices
The colours come from the model of Machado, Oliveira and Fernandes, "A Physiologically-based Model for Simulation of Color Vision Deficiency", published in IEEE Transactions on Visualization and Computer Graphics in 2009. The model works on the cones, the three kinds of photoreceptor in the eye. In dichromacy one kind is missing. The authors give a severity scale; severity 1.0 is the full loss of a cone type, which is dichromacy. This tool uses only the 1.0 values of protanopia and deuteranopia from the authors' Table 1.
Each value is a 3 by 3 matrix. The picture's red, green and blue are multiplied by it and the result is the colour seen. Each row of the matrix adds up to 1, so a grey or a white stays the same grey or white. The paper does not say which colour space the matrices use. This tool takes the sRGB picture to linear light first, applies the matrix there and converts back to sRGB; that choice is the tool's own.
What is not simulated
Tritanopia, the lack of blue cones, is not offered. Section 4 of the paper says the model "is not intended to handle tritanopia", and the values it gives for the blue cones are a shift approximation of tritanomaly, not a model of tritanopia. A tool that showed it would look more reliable than the source allows.
Milder forms of red-green deficiency (anomalous trichromacy) and complete colour blindness are not included. The model is a prediction for a typical observer; it is not a medical test, and the colours a particular person sees can differ from it.
Example: pure red under deuteranopia
Pure red is 255, 0, 0 in sRGB, which is linear light (1, 0, 0). The deuteranopia row for red gives 0.367322 for the red output, which is 163 in sRGB; the row for green gives 0.280085, which is 144. The blue output is -0.01182, which is cut to 0. The simulated red is therefore (163, 144, 0), a dark olive, and the tool writes exactly that colour into the simulated picture.
Frequently asked questions
Is my picture uploaded when I simulate it?
No. Your browser reads the picture, computes each pixel's simulated colour and writes the JPEG on your device. You can check this by turning off your internet connection after the page has loaded and simulating a small picture.
Does the simulation show exactly what a colour-blind person sees?
No. It is the model's prediction for a typical person with complete loss of one cone type. People with a deficiency differ from one another and from the model, so use it to check that a picture still works, not as a diagnosis.
Why is tritanopia not in the list?
The paper says the model is not intended to handle tritanopia, and the blue-cone values it gives are a shift approximation. The tool leaves tritanopia out rather than show a result the source does not support.
Why does the tool use linear light?
The paper gives its matrices without naming a colour space. The model describes light that the cones respond to, so the tool first converts the sRGB picture to linear light, applies the matrix and converts back. Another choice would change the colours somewhat.
Which settings should I use to check a design?
Choose protanopia and deuteranopia in turn and look for places where two colours that differ in meaning become hard to tell apart. The simulated picture shows where colour alone carries the message.
Sources
- Machado, Oliveira and Fernandes (2009), A Physiologically-based Model for Simulation of Color Vision Deficiency (paper, PDF)
- Authors' page with the severity matrices, Table 1 (read 11 November 2026)
Figures and text last checked on Wednesday, November 11, 2026.