How a Colorblind Simulator Compares a Palette

How do you check a palette for colorblindness?

Paste the colors, with a space between each one. The page draws that row as you typed it, then draws the same row seven more times: protanopia, deuteranopia, tritanopia, the three anomaly types, and achromatopsia. Each row is labeled. You do not flip a menu to see the next type, which is how a pass on one deficiency hides a fail on another.

A colorblind simulator here is a picture of your hex values, not an upload and not a vision test. This colorblind palette checker is a CVD simulator built on the Machado color vision simulation. For example, try #d32f2f #388e3c #1976d2. The red and the green land close together under deuteranopia. The blue stays apart. The note under the rows names that pair and the type. Copy keeps the hex list and that sentence.

What do protanopia, deuteranopia, and tritanopia mean?

Which cone type protanopia, deuteranopia, and tritanopia each remove is the whole label. Cone cells come in three kinds. L cones respond to longer wavelengths, M cones to the middle, S cones to shorter wavelengths. Protanopia is the simulation with no L cones. Deuteranopia is the simulation with no M cones. Tritanopia is the simulation with no S cones. The first two are the common red-green deficiencies. Tritanopia is rare. The label on the row is the type you are looking at, so a yellow swatch is not left to guesswork.

Protanomaly, deuteranomaly, and tritanomaly are the shifted-cone versions, not the missing-cone versions. This page uses severity 0.6 from the same published table, which is a milder move than severity 1.0. Anomalous trichromacy in a real eye can sit anywhere on that scale. One step is enough to see whether a palette is fragile. It is not a measurement of a person.

Achromatopsia is the row with no hue. Each color becomes a gray from its relative luminance, using the sRGB coefficients 0.2126, 0.7152, and 0.0722 on linear light. Two colors of similar lightness become similar grays even when a cone simulation still separates them. That row is there so you can see it. The collapse note does not use it, because it would name grayscale for almost every pair and hide the cone type that actually failed.

Why do pure red and pure green stay apart?

Why two mid reds and greens can collapse while pure red and pure green do not is a lightness fact, not a proverb. People say red and green look the same under deuteranopia. Pure red #ff0000 and pure green #00ff00 do not, in this model. Green is much lighter than red, and the simulation keeps that lightness. The swatches move, and they remain easy to tell apart. A proverb is a bad test color.

The pairs that collapse are the ones that already share a lightness. #d32f2f and #388e3c are a mid red and a mid green. Under deuteranopia they land close. Under protanopia they move less. Compare protanopia and deuteranopia on the same colors: both rows stay visible. A deuteranopia simulator and a protanopia color check are both on the page because those two shifts are not the same shift. If you only viewed one, the other could still fail.

Blue against yellow is the tritanopia question. #1565c0 and #f9a825 stay far apart under the red-green rows. Read the tritanopia row for that pair rather than assuming every bright pair is a red-green problem. Both distances in the note are Euclidean distance in 8-bit sRGB. One is the pair as typed. The other is the pair after the simulation. Neither is a uniform delta E.

When does the collapse note appear?

Read the note that names a collapsed pair before you change a token. Protanopia versus deuteranopia, and dichromacy versus anomalous trichromacy, are separate rows of the same Machado color vision simulation. The note is the check after the picture. It looks at every pair that started at least 40 apart in sRGB. If a cone simulation brings that pair under 36, the note names the type, both hex values, and the two distances. The closest landing wins when more than one type fails. The sentence is specific: under deuteranopia, #d32f2f and #388e3c, with the numbers the model just produced.

The note stays hidden when you typed one color, when two swatches were already the same, and when every pair that started distinct stays distinct. A quiet page is a palette that survived the six cone simulations. It is not a promise about achromatopsia, and it is not a WCAG grade. Contrast of text on a background is a different question. Open the contrast checker for the pair you will actually ship.

What does this page refuse to pretend?

It does not diagnose color vision. A simulation shows a palette to someone with typical cones, as a model of someone else. It does not sort people into types, and the anomaly rows are one severity from a table, not a threshold from a clinic. About 8 percent of men and about 0.5 percent of women have a red-green deficiency. Those rates are why the check is worth doing. They are not a result this page computes.

Blue cone monochromacy is not simulated. It is not a matrix in the Machado table this page uses, and a made-up matrix would be a confident wrong picture. Image upload is also out. A palette of hex values is the input a token file already has, and it never leaves the browser. If the colors live in a screenshot, read the hex out first.

Build the ramp in the color shades generator, then paste the stops that carry meaning: the brand, the success color, the danger color. Eight colors is the cap on this page, which is enough for a status set and too small for a whole theme. That limit keeps every simulation on one screen.

What should you open next?

Take the pair the note named into the contrast checker and read the ratio on the real background. A pair can stay distinct under deuteranopia and still fail 4.5:1 as text on white. The two checks answer different questions, and a palette needs both.

If a teammate sent hsl() or a CSS name, convert it first, then come back with hex. The simulator accepts those forms, and the copy block writes hex so the next tool sees one spelling. The arithmetic stays on the device either way.