Skip to main content

For designers, teachers and developers

Color Blindness Simulator

See any image the way a colour-blind viewer sees it — with severity you can dial, not just the complete forms.

Loading the simulator…

Short answer

A colour blindness simulator shows a normally-sighted viewer what an image looks like to someone with a colour vision deficiency. This one applies the Machado, Oliveira & Fernandes (2009) physiological model to your image entirely in the browser, covering protan (red-weak), deutan (green-weak) and tritan (blue-yellow) at any severity from 0 to 100%.

Severity matters: most people with a deficiency are anomalous trichromats with partial loss, so a simulator fixed at complete dichromacy overstates what they actually experience. Nothing is uploaded — the image is processed locally and never leaves your device.

Model
Machado 2009
Physiologically-based, the current standard
Severity
0–100%
Continuous, not just complete dichromacy
Your image
Stays local
Processed in-browser, never uploaded

How to use the simulator

1. Pick an image

Start with a built-in scene — a chart whose series differ only by hue, colour-coded status UI, a hue wheel, or signal colours — or upload your own design, screenshot or photo.

2. Choose type and severity

Protan, deutan or tritan, then set severity. Deutan around 40–60% is a good default for 'the most common real case'; 100% is the complete form, which is far rarer than people assume.

3. Drag the divider

The split view puts the original and the simulation edge to edge, so a difference you would rationalise away in two separate images becomes obvious. Save the frame with its settings stamped on it.

Which simulation model is this, and why?

Simulators disagree with each other, and the disagreement is not cosmetic — it changes how severe the condition looks. This one states its method: the Machado, Oliveira & Fernandes (2009) physiological model, with severity interpolated between the anchors published in that paper.

Underlying model

This simulator
Machado, Oliveira & Fernandes (2009) — derived from a physiological model of shifted cone photopigments
Commonly found elsewhere
Older simulators often use the 1990s LMS-projection approach, which was built for the complete forms only

Severity

This simulator
Continuous 0–100%, interpolated between the paper's published anchors
Commonly found elsewhere
Frequently a single fixed setting: full dichromacy

What that changes

This simulator
You can show anomalous trichromacy — the common case, roughly three quarters of red-green deficiency
Commonly found elsewhere
Everything looks like the most severe form, which overstates the impairment

Where the image goes

This simulator
Nowhere — WebGL in your browser
Commonly found elsewhere
Often a server upload

What a simulation can and cannot tell you

  • Does this colour pairing survive? Yes — if two elements become hard to tell apart in the simulation, that is a real risk for a real audience. This is what the tool is for.

  • How bad is it for a given person? An approximation. Severity varies between individuals with the same diagnosis, and your screen is not calibrated, so read the result as a class of experience rather than one person's.

  • Does it show what they actually perceive? No. A lifetime of adaptation — using lightness, texture and context — is not reproducible by an image transform. Simulations tend to look more disabling than lived experience is.

  • Checking your own colour vision? This is the wrong tool — you would be viewing it through your own vision. Take the colour blind test instead.

How accurate is a colour blindness simulator?

Accurate enough to make design decisions with, and not accurate enough to describe one person's experience. The Machado 2009 model is derived from the physiology of shifted cone photopigments rather than fitted to appearance judgements, and it reproduces the published dichromatic results at full severity while remaining well-behaved in between. That is what makes it the right basis for checking whether a chart, a map or an interface still works.

Three limits are worth stating plainly. Your screen is not calibrated, so the input colours the model receives are already a little off. Colour vision deficiency varies between individuals with the same clinical label, so no single matrix speaks for everyone. And a simulation cannot reproduce a lifetime of adaptation — people who have never had full colour vision use lightness, texture, context and learned associations in ways a static image transform does not capture.

The practical reading: if something becomes ambiguous in the simulation, it is a real risk worth fixing. If it survives, you have evidence but not proof — the robust fix is never to rely on hue alone in the first place.

Why does severity matter more than type?

Because most colour vision deficiency is partial. The complete forms — protanopia, deuteranopia — are the ones every illustration shows, but they are the minority. The far more common condition is anomalous trichromacy: all three cone types are present, one of them is shifted, and colour discrimination is reduced rather than absent. Someone with mild deuteranomaly may go decades without noticing.

This has a direct consequence for design review. A simulator locked at 100% tells you what a small share of your audience sees and quietly implies everyone else sees the same thing. Sweeping severity from mild to complete tells you something far more useful: the severity at which your design stops working. If a chart falls apart at 40%, that affects a large population. If it only fails at 100%, the exposure is different — and so is the priority.

That is why the slider here is continuous and why the exported frame stamps the severity onto the image. A simulation without its settings is not evidence of anything once it leaves the screen.

What should I change when a design fails the simulation?

Add a second channel. WCAG success criterion 1.4.1 (Use of Color, Level A) requires that colour is never the only visual means of conveying information — and almost every failure this simulator surfaces is a 1.4.1 failure. Direct-label chart series instead of relying on a legend; add shape or dash patterns to lines and markers; pair status colours with an icon or a word; put a texture or a pattern on map regions.

Choosing 'colour-blind-safe' palettes helps but does not finish the job. Palettes that separate cleanly for deutan can still collapse for protan or tritan, and any palette degrades once elements get small, thin or semi-transparent — a 1px line and a large filled block with the same colour are not equally distinguishable. Check the actual artefact at the actual size, which is what this tool is for.

The ordering that tends to work: fix the information channel first, then tune the palette, then verify with the simulator across all three types at a mid severity as well as at 100%.

Colour blindness simulator — frequently asked questions

What is a colour blindness simulator?

A tool that transforms an image to approximate how it appears to someone with a colour vision deficiency. It is used for accessibility review — checking that charts, interfaces, maps and signage stay readable — and for teaching, since it lets a normally-sighted person see the problem rather than be told about it.

Which model does this simulator use?

The Machado, Oliveira & Fernandes (2009) physiologically-based model, published in IEEE Transactions on Visualization and Computer Graphics. It derives simulation matrices from shifted cone photopigment spectra, which is why it behaves sensibly at partial severities rather than only at the complete dichromatic forms.

Is my image uploaded anywhere?

No. The image is decoded and transformed in your browser using WebGL, with a canvas fallback. It is never sent to a server, which also means you can safely simulate unreleased designs and internal screenshots.

Can this simulator tell me if I am colour blind?

No — it works the other way around. A simulator shows a normally-sighted viewer what a deficiency looks like; it cannot assess your own vision, because you would be viewing the simulation through whatever colour vision you already have. To check your own, take a colour blindness test instead.

Which type should I test against?

All three, but start with deutan: green-weak deficiency is the most common form by a wide margin. Protan matters especially where reds must stay visible, since protans also see reds as darker. Tritan is rare and usually acquired, but blue-yellow palettes are common enough in data visualisation to be worth a check.

Why does my design look fine at 100% severity?

Some palettes genuinely survive, but check whether you are relying on lightness rather than hue — that is a legitimate and robust strategy, and the simulator confirms it works. If instead the elements are large, high-contrast blocks, try the same palette at the real size: thin lines, small text and 1px borders lose distinguishability far sooner than filled shapes.

Can I use these simulations in a presentation or paper?

Yes. Saved frames carry the simulated type and severity in the caption so the image stays interpretable on its own, and the underlying model is published and citable. Please cite Machado, Oliveira & Fernandes (2009) for the model itself.

Keep reading

Color blind test

Checking your own colour vision, rather than simulating someone else's.

Color blind filter

The inverse tool: a live camera filter that assists a colour-blind viewer.

Types of color blindness

What protan, deutan and tritan mean, and how common each one is.

Accessibility

How this site handles colour, contrast and non-colour cues.

Sources

Last reviewed August 10, 2026. Simulation is an approximation for design and teaching purposes — it is not a clinical assessment of any individual's colour vision.