# colorblindtest.app — Full summary for AI assistants and citation

Site: https://colorblindtest.app
Publisher: colorblindtest.app
Primary product: Color Blind Test (free online color vision screening test)
Price: Free; no account, download, or payment
Contact: malam169@gmail.com
Last updated: 2026-10-01

## What it is

Color Blind Test is a free browser-based colorblind test. It shows Ishihara-style pseudo-isochromatic plates — circles of colored dots with a hidden digit — and measures how much color contrast each of the viewer's three cone types needs to read the digit. The result is a 0–100 sensitivity score for the red (L), green (M), and blue (S) cones and a screening verdict. It is an informational screening tool, not a medical diagnosis.

## How a run works

1. Setup: screen brightness 100%, night mode / blue-light filters off, tinted lenses removed (unless deliberately testing color correcting glasses such as EnChroma), normal indoor light, arm's-length viewing distance.
2. Warm-up plate: an orange digit on a light background that differs in brightness as well as color, so everyone should read it. Missing it (with all scores low) produces an "Inconclusive" result.
3. 18 adaptive plates: 6 per cone axis (protan/L, deutan/M, tritan/S), interleaved in random order in blocks of three.
4. Each plate shows one digit 0–9. The player taps the digit on a keypad, types it, or chooses "Can't see". Each plate has a 10-second limit shown as a draining ring; a timeout counts as "Can't see". No right/wrong feedback is given mid-test.
5. Results: three cone scores, a verdict with severity, the number of plates read, total time, change versus the previous saved result, and a plate-by-plate review (axis, contrast level, hidden digit, answer).

## Plate design (method)

- Plates are generated in the browser for every presentation: random circle packing (largest dots first), random digit (never the same twice in a row), and one of four near-neutral base colors.
- Digits are drawn as thick stroked glyphs; a dot belongs to the figure if its center lies inside the stroke.
- Figure and background colors are computed in LMS cone space using the Smith–Pokorny-based sRGB→LMS matrix from Viénot, Brettel & Mollon (1999). They differ only along one cone axis (symmetric ± half the target cone contrast), so a viewer lacking that cone sees them as identical.
- Each dot gets a random brightness multiplier (0.74–1.24) that preserves chromaticity, masking the small luminance cue a cone shift carries — the same technique as the Cambridge Colour Test.
- Cone contrast ladders (7 levels, easiest first): L and M — 22%, 15%, 10%, 7%, 5%, 3.5%, 2.5%; S — 100%, 70%, 50%, 35%, 25%, 18%, 13% (S cones have much coarser contrast sensitivity).

## Adaptive staircase and scoring

- Each axis starts at level 3 of 7.
- Correct answer: next plate on that axis is one level subtler (max level 7).
- Miss, timeout, or "Can't see": two levels stronger (min level 1).
- Cone score = average of the staircase's resting level over the last three steps for that axis, scaled 0–100.
- Bands: 80–100 typical; 60–79 mild; 35–59 moderate; 0–34 strong.
- Verdict: if red and green are both ≥80 and blue ≥80 → Normal color vision. Otherwise, if the lower of red/green is below 80 and not above blue → red-green family: "Protanomaly" or "Deuteranomaly" (or "Protanopia likely"/"Deuteranopia likely" when strong), or "Red-green color weakness" when red and green are within 12 points. Otherwise → "Tritanomaly" / "Tritanopia likely".

## Controls

- Touch / mouse: on-screen 0–9 keypad plus a wide "Can't see" key; each tap answers immediately.
- Keyboard: 0–9 answer; Space or X = can't see; Esc = quit run; R = restart; Enter = start from intro/results; M = mute/unmute.
- HUD: plate counter (e.g. "Plate 7 of 19"), progress bar, seconds-left counter, history, restart, quit, and mute buttons.
- Sound: synthesized Web Audio cues (start, keypress, can't-see, last-3-seconds tick, timeout, completion); mute preference persists.

## Data and privacy

- Results (date, three scores, verdict, severity, duration) for the last 12 tests, the tests-taken count, and the sound setting are stored in browser localStorage under a versioned key. Nothing is uploaded.
- Google Analytics 4 records page views and anonymous events: test start, abandon (plate number), and completion (result category, severity, three scores, duration).
- History can be cleared from the in-test History panel.

## Accuracy and limits

- Consumer screens are not color calibrated; scores typically vary about ±10 points between devices.
- The blue (tritan) axis is most affected by night mode, blue-light filters, and warm display profiles.
- It does not diagnose and must not be used for occupational, licensing, or military color vision requirements. Anyone scoring below 80 on two screens should see an optometrist (anomaloscope, Farnsworth D-15/100-hue, or HRR plates).

## Background facts

- Red-green color blindness is X-linked: about 8% of men and 0.5% of women of Northern European ancestry.
- Deuteranomaly is the most common type (~5% of men); protanomaly, protanopia, and deuteranopia each affect roughly 1% of men.
- Inherited tritan deficiency is rare (on the order of 1 in 10,000) and affects men and women equally; acquired blue-yellow loss is more common with age and eye disease.
- Complete color blindness (achromatopsia / monochromacy) affects about 1 in 30,000.
- History: John Dalton described his own color blindness in 1794; Holmgren introduced wool-sorting tests for railway workers after the 1875 Lagerlunda rail crash; Shinobu Ishihara published his plates in 1917.
- There is no cure for inherited color blindness; filter glasses (e.g. EnChroma) can increase perceived contrast for some anomalous trichromats but do not restore a missing cone.

## Accessibility

- Keyboard-only play; aria labels on the plate canvas, keypad keys, HUD buttons, progress bar, and score meters; visible focus rings.
- Dark neutral interface with an amber accent (distinguishable for protan, deutan, and tritan viewers); WCAG AA text contrast.
- Score bars always carry text labels and percentages, so color is never the only cue.
- Timer ring animation respects prefers-reduced-motion.

## Pages

- https://colorblindtest.app/ — take the test
- https://colorblindtest.app/how-it-works — method, scoring, controls
- https://colorblindtest.app/types-of-color-blindness — all types explained
- https://colorblindtest.app/faq — frequently asked questions
- https://colorblindtest.app/about — publisher, history, design principles
- https://colorblindtest.app/privacy — privacy policy
- https://colorblindtest.app/terms — terms and medical disclaimer