Color blindness simulator that names the pairs that collide

Deuteranomaly, then deuteranopia — the green-side conditions are the common ones. Around 8% of men of northern-European descent and 0.5% of women have some inherited color-vision deficiency, and roughly 6 in 100 men are on the deuteran side, split about five to one between the anomalous and the dichromatic form. Protan conditions run at about 2 in 100 men, tritan defects are rarer than 1 in 10,000 and hit both sexes equally because the gene sits on chromosome 7 rather than the X, and achromatopsia is around 1 in 30,000. If you only ever check one column, check the deuteranopia one.

Because a missing cone is not a missing channel. Zeroing red would turn #ff0000 into black; a protanope's eye still receives the light, it just cannot separate it from the green signal, so the color is projected onto the plane the two remaining cone types can span. Pure red comes back as #6d5f00 for protanopia and #a39000 for deuteranopia — an olive in both cases, but a much darker one for the protanope, because the L cone carries most of red's luminance and the M cone does not. That luminance difference is why protanopes are the group that loses brake lights and red text on dark backgrounds.

No — it is what a person with typical vision would have to see to face the same discrimination problem. Every dichromatic model in circulation, this one included, is anchored to reports from unilateral dichromats: the handful of documented people with one typical eye and one affected eye, the only population that can describe both views and say which colors match. It says nothing about a lifetime of adaptation: someone born deuteranopic has learned which olive means ripe and calls plenty of things red with perfect confidence. Treat the output as a discrimination test — can these two be told apart — rather than a photograph of anyone's experience.

Contrast and hue discrimination are two different measurements and neither predicts the other. Tailwind's red-500 and green-500 sit 127 ΔE apart as authored and 12.7 apart under deuteranopia — still visible side by side, hopeless as two lines on a chart — while both keep exactly the contrast ratio against white that they had. Run the pair through the color contrast checker and nothing looks wrong, because that formula compares light output and is blind to hue by design.

It changes the real ratio, and WCAG deliberately does not ask you to check it. Pure red on black measures 5.25:1 as authored and 3.29:1 through the protanopia simulation, so an AA pass becomes a fail; red #dc2626 on white drops from 4.83:1 to 4.00:1 under deuteranopia. Success Criterion 1.4.3 is measured on the colors you shipped, so those simulated numbers are not a conformance failure — but 1.4.1 Use of Color is a criterion, and it is the one that a red-versus-green status dot breaks.

Not here — this page takes color values, and simulating an image is a different job with a different tool. Chrome and Edge do it live under DevTools → Rendering → Emulate vision deficiencies, and Firefox has the same thing in the Accessibility panel, both on the page as rendered rather than on a static capture. If what you have is a picture, lift the colors out of it first with the image color picker or pull the dominant palette with color from image, then bring those hex codes back here where the pair-by-pair report lives.

Because ΔE 12 is a comfortable difference for two swatches touching each other and a poor one for two areas 300 pixels apart. The list flags anything under 15 and marks anything under 2.3 — the just-noticeable difference — as having merged completely. CIE76 is the metric here, which over-penalises differences among saturated colors and under-reports them among the desaturated olives and greys that dichromatic simulation produces, so read the number as a ranking of your riskiest pairs rather than as a threshold to argue over.