Color Blind Test

A pseudoisochromatic plate — the dotted circles used in colour vision screening — works by removing every cue except hue. The figure and the background are built from dots of randomly varying size and lightness, so there is no edge to trace, no brightness step, and no shape to infer. The only thing separating the number from the field is colour.

The trick is choosing that colour. Human colour vision starts with three cone types: L (long wavelength, “red”), M (medium, “green”) and S (short, “blue”). When one is missing or shifted, whole sets of colours that a trichromat separates easily collapse onto each other. Those sets trace out confusion lines in colour space.

Put the figure and the background on the same confusion line, matched in lightness, and the result is a plate that reads as a clear number to one observer and as an unbroken field of dots to another.

A clinical plate test is a physical object viewed under a specified illuminant, and it is scored by someone watching how you respond. None of that survives the trip to a web page:

• Every panel renders colour differently. Two phones side by side will not show the same salmon. A plate tuned to vanish on one screen can stay faintly visible on another.
• Night mode destroys the test. Blue light filters warm the whole display, which shifts exactly the hues under examination.
• Ambient light shifts perception. The same screen read under a tungsten lamp and under daylight gives different answers.
• Screening cannot grade severity. Plates separate “likely deficient” from “likely not”. They do not measure how much, and they do not reliably tell protan from deutan — clinically that needs an anomaloscope, an instrument where you match a yellow against a red-green mixture.

This is why the test above reports an axis and a pattern rather than a named diagnosis. A result of “red–green pattern detected” is a reason to book an eye test, not a label to adopt.

The plates here are generated in your browser rather than scanned from a book. Dot positions come from a seeded random packing, so a given plate redraws identically every time, and a glyph mask decides which dots take the figure palette.

The palettes themselves were not picked by eye. Each candidate pair is run through a dichromat simulation based on Viénot, Brettel & Mollon (1999): convert sRGB into LMS cone space, collapse the missing cone onto the plane spanned by the remaining two, convert back. A pair only ships if the figure sits far from the background for a trichromat and lands on top of it once simulated.

That check matters more than it sounds. The obvious choice — orange dots on green dots — fails: hand-picked pairs almost always differ in lightness, and lightness survives the simulation, leaving the number readable to everyone. The palettes below are the measured gaps for the set actually in use, on a weighted RGB distance where higher means more separable.

If nothing was detected, there is nothing to act on. Plate tests are good at ruling out the common inherited types.

If a red–green pattern showed up, it is worth confirming with an optometrist, particularly before anything that depends on it. Aviation, rail, some maritime and electrical roles, and parts of the armed forces apply colour vision standards, and they test with their own approved method — an online result carries no weight there.

If your colour vision has changed recently, treat that differently from a lifelong pattern. Inherited deficiency is stable from birth and affects both eyes equally. Colour vision that shifts over weeks or months, or that differs between your two eyes, is acquired, and can relate to the optic nerve, the retina, or medication. Get that looked at rather than logged as a test score.