Colour Vision Self-Check

Features

• Eight plates — one control, four red-green, three blue-yellow — all drawn on this page
• The control differs only in brightness, never in hue, and missing it discards the whole run
• Red-green and blue-yellow pairs are luminance-matched (contrast < 1.2), so hue is the only cue
• Dot positions are random per plate, so a second run is a different pattern
• Accepts full-width digits, spaces and commas
• A discarded run lists no answers — when the conditions failed, comparing answers proves nothing
• Reports only how many plates you read, with no “mild / moderate” grading (that needs a clinical scale)
• Runs locally: the answers are compared in your browser and never uploaded

How to use

1. Read at a normal distance — do not lean in — and give it about a second
2. Type the digits if you can read them; press “cannot see it” if you cannot (guessing defeats the purpose)
3. After eight plates, read the result; if the control was missed, the run is void
4. Change the setting or the screen brightness and try again — differing results between runs are normal

FAQ

Does this mean I am colour blind?

No — this tool cannot diagnose anything. Three reasons: the plates are drawn here rather than being the clinical standard ones (the real Ishihara plates are copyrighted and cannot be reproduced); the colours depend on your screen, which is not calibrated; and colour vision deficiency itself comes in several types and degrees, assessed by an optometrist with standard plates under standard lighting. So the strongest statement is “you could not read these plates” — and if that keeps happening, it is worth a proper colour vision test. Colour vision deficiency is common and affects very little of daily life.

What is the first “control plate” for?

It checks whether the conditions for this run hold at all. Its two colours differ only in brightness, never in hue, so with a normally lit screen and a moment’s attention it is readable by anyone, colour vision aside. Missing it therefore points at the conditions — a dim screen, glare, bright ambient light, no glasses — rather than at colour vision. In that case the entire result is discarded and said so plainly, instead of being counted as “one plate wrong”. The greatest risk for a self-check is presenting an invalid measurement as a diagnosis.

Why does it matter that the two colours are “luminance-matched”?

It is the crux of whether such a plate works at all. If the two colours differed in brightness as well as hue, a colour-deficient viewer could still read the digits from the brightness contrast alone — the plate would have degraded into an ordinary greyscale image that tests nothing. So in the red-green and blue-yellow plates the two relative luminances are aligned to match (the second colour’s lightness is found by binary search, and a test pins the contrast below 1.2), leaving hue as the only difference.

Why do you sometimes not show me the answers?

Because when the control plate was missed the whole run is already void, and listing the answers would only invite you to check them one by one — while that comparison proves nothing, since the conditions themselves failed. Fix the setting until the control plate reads, run it again, and the plates you then miss are listed with their answers.