How this site works
What we build, how we decide what to claim, and where every fact comes from.
In short
Color Vision Labs builds free color vision screening and simulation tools, and publishes sourced reference material on color vision. It is not a clinic, and nothing here is a diagnosis.
What follows is how the site decides what it will and will not claim — which is the part worth judging it on.
Every online color vision tool is limited by the same thing: an uncalibrated display, in unknown lighting, viewed at an unknown distance. That does not make screening useless — it makes it honest to say which questions survive those conditions and which do not.
The last row is the one that matters most. A tool that answers every question asked of it is a tool that is guessing at some of them, and the guesses are not labelled. We would rather publish the shape of the gap.
Every session draws new plates from a seeded generator rather than serving a fixed set of images. Answers cannot be memorized, no answer key can circulate, and someone retesting a week later is not re-reading a picture they have already seen. This matters most where screening is used as a gate — a test you can rehearse is not a test.
Plates carry randomized lightness variation, so a digit can only be found by its color, not by being lighter or darker than its surroundings. Without that noise, a screening image can be solved on brightness cues, which is exactly how a static plate ends up passing someone it should not.
The Farnsworth D-15 is scored with the Vingrys & King-Smith confusion-angle method — the same geometry used on the paper test — rather than a pass/fail cap count, so the result names an axis instead of a verdict.
Simulated views use Machado, Oliveira & Fernandes (2009) with severity as a continuous parameter. Most simulators render only complete dichromacy, which overstates what the majority of color-blind viewers experience: anomalous trichromacy — a partial shift — is the common case, and a design judged only against the extreme gets compensated in the wrong direction.
Any claim that could change a decision — licensing rules, occupational standards, prevalence figures — links to the body that sets it, not to a summary of it. Pages carry the date they were last reviewed, data tables are re-checked on a schedule, and corrections are made in place with the review date moved forward rather than quietly edited.
The standards bodies cited most often across the site are the National Eye Institute, American Academy of Ophthalmology, NHS, the FAA and the national transport authorities that set driver-licensing rules — each cited on the specific page where its rule applies.
These are screening and self-assessment tools. Anything official — a licence, a medical, a job requirement — needs an eye-care professional with an anomaloscope or a CAD test. Where a page touches those rules, it says what the rule is and links to the authority that sets it, and stops there.
Tests run in your browser. Camera frames and photos in the filter tool are processed on-device and are never uploaded. Results are stored anonymously, without anything that identifies a person.
This site does not put an expert's credentials at the top of a page in place of evidence. Every claim links to the body that sets the standard, carries the date it was last checked, and states plainly where a screen falls short of a clinic. Check the sources instead of trusting the author — for a health page that is the stronger test, and it is the one we would rather be held to.
If something here is wrong, out of date, or overstated, write to hello@colorvisionlabs.com. Corrections are made in the page itself and the review date moves with them. Reports that a rule has changed are especially welcome — licensing and occupational standards are exactly the kind of fact that goes stale quietly.
This page was last reviewed August 11, 2026. Accessibility commitments and known limits are documented separately in our accessibility statement.
Color blind test
Generated plates, an estimated type, and a stated limit on severity.
Color blind filter & simulator
Machado 2009 with continuous severity, running entirely on your device.
Driving licence rules by country
A sourced table with the governing rule and authority on every row.
Tetrachromacy test
The page we wrote instead of a test a screen cannot honestly run.