Simulate color blindness on an image

This tool shows one image four times — as it is, and as it appears to viewers with protanopia, deuteranopia and tritanopia — so you can tell whether a chart, a map or an interface still works when a class of cone is missing. The simulation uses the Viénot, Brettel and Mollon matrices applied in linear light: every pixel is taken out of the sRGB gamma curve, mixed, and re-encoded, because multiplying gamma-encoded bytes gives visibly wrong colours. A severity slider blends between ordinary colour vision and full dichromacy, and a difference view replaces each panel with a greyscale map of how far its colours travelled.

How it works

  1. Drop an image Drop or browse for one image. Four panels appear: the original, then protanopia, deuteranopia and tritanopia, each labelled with what is missing and roughly how many people it affects.
  2. Set the severity The slider runs from 0 to 100. Zero is ordinary colour vision and all four panels match; 100 is full dichromacy. The milder anomalous forms, which are far more common, sit in between.
  3. Turn on Show what moved This replaces each simulation with a greyscale map of how far its colours travelled. Black did not move; the brighter a pixel, the further it went. Each panel also reports the share of visible pixels that shifted more than a just-noticeable step.
  4. Choose the output format PNG is lossless and the safe choice for a chart. WebP and JPEG are also offered, with a quality slider for the lossy ones.
  5. Save the panels The download button on a panel writes that one at the image's full size. Download all four as ZIP writes the set. Downloads follow what is on screen, so with the difference view on you get difference maps rather than simulations.

Questions

Which types of colour blindness does it simulate?
Three: protanopia (no red cones), deuteranopia (no green cones) and tritanopia (no blue cones). Protanopia and deuteranopia both collapse the red-green axis, so a red-green chart fails in those two panels; tritanopia leaves red and green alone and flattens blue against yellow instead.
Is this a test for colour blindness?
No. It simulates how an image looks to someone with a colour vision deficiency, using published dichromat matrices, so that a designer can check a chart or an interface. It diagnoses nothing about the person looking at it, and the severity slider is a blend toward full dichromacy rather than a clinical measurement.
What does the "% moved" figure mean?
It is the share of visible pixels whose colour shifted by more than a just-noticeable step in that simulation, and it is measured on the on-screen preview rather than on the full-size image. Fully transparent pixels are excluded, and their alpha is never rewritten by the simulation.
Are the four panels uploaded anywhere to be generated?
No. The image is decoded, simulated and encoded in your own browser tab, and the panels you download are written there too. Above about 8 megapixels the tool warns that writing four full-size panels takes a moment — all of which happens on your machine.

Every tool