Testar daltonismo
Color Blindness Test shows eight plates of coloured dots with a number hidden in them, where the number is separated from its background only by colour and not by brightness. The plates are generated here rather than scanned from the Ishihara set, and their colours are chosen by simulating each deficiency and keeping the pairs that collapse under it. It is a screening indication on an uncalibrated screen, not a diagnosis.
A interface desta ferramenta está em inglês.
The plate test
Eight plates. Type the number you see, or leave it empty if you cannot see one. There is no time limit and no score to beat — an honest answer is the only useful one.
Sit at a normal distance in ordinary room light, with your screen at its usual brightness and any night-shift or blue-light filter turned off — those change the colours the test depends on.
See it the way they do
Paste a colour, or drop a screenshot of your chart or interface. Everything is decoded and transformed in this tab.
O guia abaixo está disponível apenas em inglês.
How does Color Blindness Test work?
A pseudo-isochromatic plate hides a figure in a field of dots where the figure and the ground differ only in hue. Someone whose colour vision separates those two hues reads the number immediately; someone whose does not sees an even field. The dots vary in size and tone so the figure cannot be found by shape, edge or brightness — which is what makes it a test of colour rather than of eyesight.
Why these plates are generated rather than scanned
The Ishihara plates are a specific copyrighted set from 1917, and the scans circulating online have been through a scanner, a compression pass and whatever colour profile the page was saved with. By the time one reaches your screen the exact hues it depends on have moved, which is most of the reason online colour-vision tests disagree with each other. Nothing here is reproduced from that set.
How the colours are chosen
Not from a table. For each deficiency the page searches hue pairs and keeps the one that is furthest apart in OKLab before simulation and closest together after it, using the same Machado matrices the simulator below uses. A pair qualifies only if the two colours land on top of each other for that deficiency.
That search is why protanopia and deuteranopia get different plates. Both are red-green deficiencies and it is tempting to treat them as one, but the classic pink-and-green deutan pair stays clearly separable under protanopia — so a test using one pair for “red-green” would miss protans entirely. The pair it finds for protanopia is a violet and a blue, which looks nothing like what you would expect and is exactly what the simulation says.
The control plates
The first and last plates use colours nobody confuses. They are not filler: if those are missed, the problem is the screen, the room light or the instructions, and the rest of the result should be thrown away rather than reported. Any test without them is reporting confidently on data it has not checked.
What this cannot tell you
Severity, and the difference between full dichromacy and the far more common anomalous trichromacy. It also cannot correct for your screen: uncalibrated displays vary enough that a borderline result may be about the monitor. A clinical diagnosis uses an anomaloscope under controlled lighting, and if the result here matters to you — for a licence, a job, a medical question — that is the test to ask for.
What you see
8 plates 2 controls everyone reads 4 red-green 2 blue-yellow
What it indicates
controls 2 / 2 read red-green 4 / 4 read blue-yellow 2 / 2 read No indication of a deficiency.
What options and edge cases does Color Blindness Test support?
| Parameter | Type | Default | Behaviour & edge cases |
|---|---|---|---|
| Deuteranomaly | green, reduced | ≈5% of men | The most common form by a wide margin. The green cone is shifted rather than missing, so reds and greens are muddled but not identical. |
| Deuteranopia | green, absent | ≈1.1% of men | No green cone at all. Red and green collapse to a shared yellow-brown. |
| Protanopia | red, absent | ≈1.0% of men | No red cone. Reds darken as well as shifting, so red text on a dark background can disappear entirely. |
| Tritanopia | blue, absent | ≈0.01% | No blue cone. Blue and green converge and yellow reads as pink. Not sex-linked, so it affects men and women equally. |
| Achromatopsia | no colour | ≈0.003% | Complete absence of colour vision. The honest test of whether a design still works on lightness alone. |
| Overall | — | 8% / 0.5% | About one man in twelve and one woman in two hundred has some form. The genes for the red and green cones are on the X chromosome, which is why. |
| Plates here | 8 | 2 controls | Four red-green and two blue-yellow, bracketed by two plates everyone should read. |
| Diagnosis | — | anomaloscope | A clinical test under controlled lighting. Nothing on an uncalibrated screen replaces it. |
Frequently asked questions
How accurate is an online colour blindness test?
Good enough to tell you whether to see an optometrist, and not good enough for anything else. Your screen is not calibrated, the ambient light is unknown, and a borderline result may be about the monitor rather than your eyes. This test at least generates its plates from the actual simulation rather than using a scanned image that has been through compression and a colour profile, and it brackets the result with control plates so an obviously invalid run can be discarded. It still cannot tell you severity, and it cannot distinguish full dichromacy from the much more common anomalous form.
What does it mean if I miss the red-green plates but read the rest?
It is consistent with a red-green deficiency — deuteranomaly or deuteranopia if the greens confuse, protan if the reds also look darker than expected. That is about one man in twelve. It is worth confirming with an optometrist, particularly if you work with colour-coded information or are applying for a licence that tests for it. It is also worth knowing that most people with it navigate the world without difficulty and discover it in a test like this one.
Why are your plates different from the Ishihara ones I have seen?
Because these are generated, not reproduced. The Ishihara set is copyrighted, and the scans circulating online have lost the colour accuracy the test depends on. These plates pick their colours by simulating each deficiency and keeping pairs that collapse under it, which means they are correct for the simulation rather than correct in 1917 and degraded since. A side effect: the protanopia plates use a violet and a blue rather than the red and green you would expect, because that is the pair the simulation says protanopia cannot separate.
Can colour blindness be cured or corrected?
No, and the tinted glasses sold for it do not correct it either — they increase the separation between some confusable colours at the cost of distorting others, which some people find useful and many do not. Nothing restores a missing cone type. Gene therapy has restored colour vision in animal models and is not a clinical treatment for humans.
How do I design something that works for colour-blind users?
Never let colour be the only carrier of meaning. A red and green status dot becomes two identical dots for one man in twelve; add a shape, an icon or a word and it works for everyone. Keep contrast high, because deficiency reduces the separation you have to work with. And check your work with the simulator below — pasting your own chart or interface into it is faster than reasoning about it, and considerably more convincing.
Does this send my answers or my image anywhere?
No. The plates are drawn in this tab, your answers are held in the page and forgotten when you close it, and an image dropped into the simulator is decoded by your own browser. There is no server here to receive any of it.