Colour Threshold Trainerdemo

How a computerised colour-vision threshold test works: a browser demo

This page rebuilds the procedure of the Colour Assessment and Diagnosis (CAD) test as it is described in published research. You can see and try out how such a test runs: a coloured square hidden in flickering brightness noise, and a staircase that keeps lowering the colour until you can no longer see it.

Please read first

  • This is a demonstration, not a medical test. It does not diagnose, screen for or rule out any colour vision deficiency. Do not use it for medical, aviation, occupational or any other decisions.
  • It is derived from the literature, not from the original test. Several values had to be reconstructed or assumed. Nothing here has been validated against the real CAD test or any clinical test.
  • Your screen is not calibrated. The numbers shown cannot be compared with official results and say nothing about how you would do in an official test.
  • Not affiliated with or endorsed by City, University of London, City Occupational Ltd, EASA, the UK CAA or any other authority. "CAD" is only used to name the test the literature describes.
  • If you have questions about your colour vision, ask an eye-care professional or an aeromedical examiner.

What you'll do

  1. Set up once: match a bank card on the screen and enter your viewing distance, so the pattern has the size the literature describes.
  2. Learn: ten easy trials to get used to the four corner keys.
  3. Run: after each ¾-second presentation and a beep, press the corner where the coloured square ended. A full run takes about 10 minutes.

Your data stays on this device

There is no account on a server and no cookie. Profiles, setup and runs are kept only in this browser's storage. You can export them as a backup file. Clearing your browser data deletes them.

Anonymous usage statistics are counted with a self-hosted, cookieless Umami instance: page views and events such as "run finished". Names, PINs and scores are never sent.

Colour Threshold Trainerdemo

Who is practising?

Profiles exist only in this browser on this device. Each one keeps its own screen setup and run history.

New profile

The PIN only keeps profiles apart on a shared device. The data is not encrypted.

Colour Threshold Trainerdemo

This screen looks different from your setup. The pattern size and colours may be wrong until you redo the setup.
No setup yet for this profile. Measure your screen and set your viewing distance first, otherwise the pattern has the wrong size.

Learn

≈ 30 s

Ten strongly coloured targets to get used to the four corner keys and the beep.

Full run

≈ 10 min

All 16 colour directions described in the literature (12 red-green, 4 yellow-blue), each with its own interleaved staircase.

Red-green run

≈ 7 min

Only the 12 red-green directions, as in the shorter red-green procedure the literature describes.

How a run works

  1. The screen turns grey. Look at the small cross in the centre and keep your eyes there.
  2. A square of flickering checks appears. A coloured square moves diagonally through it for about ¾ s.
  3. After the beep, press the corner where the square ended. If you're not sure, guess.
  4. The colour gets weaker after two correct answers and stronger after a mistake. Don't follow the square with your eyes.
79 5 13

Numpad 7 9 1 3, the letter corners Q E Y/Z C, or click the matching quarter of the screen.
R shows the last presentation again. Space pauses. Esc stops.

1 Screen size

Hold a bank card (ISO ID-1, 85.6 × 54 mm) against the screen. Drag the slider until the rectangle is exactly the size of the card.

bank card

2 Viewing distance

The literature describes 1.4 m. The pattern is scaled to whatever distance you enter so it keeps the described visual angle. Measure from your eyes to the screen.

3 Colour space

Mac and iPhone screens cover Display P3, which leaves room for stronger colours. Use sRGB on ordinary monitors.

4 Before each run

What this demo rebuilds

The Colour Assessment and Diagnosis (CAD) test was developed by John Barbur at City, University of London. According to the literature, it measures how much colour a person needs to see a target when brightness cues are masked. The target has the same luminance as the background, and the checks around it flicker randomly in brightness. Thresholds are measured along 16 directions in the CIE 1931 colour diagram and reported in Standard Normal (SN) units, where 1 SN is the median threshold of a large group of young observers with normal colour vision.

This demo follows those published descriptions. It is independent work and does not use the original software, data or materials.

ParameterDescribed in the literatureThis demo
BackgroundCIE (0.305, 0.323), 24 cd/m²Same chromaticity; luminance relative to your screen
Noise field15 × 15 checks, ≈ 3°Same, scaled to your distance
Target5 × 5 checks, ≈ 1°Same
Luminance noise±45 %, each check changes every 50–80 msSame
MotionDiagonal, ≈ 3°/s through ≈ 2.26°, 4 directionsSame (≈ 0.75 s)
Colours12 red-green (140–180°, 320–355°) + 4 yellow-blue (60, 64, 240, 244°)Same
ProcedureFour-alternative forced choice, interleaved staircases, beep after each presentationSame principle; step sizes assumed
UnitsSN units along 334° (red-green) and 67° (yellow-blue)1 SN reconstructed from published cone contrasts
DisplaySpectrally and photometrically calibrated, dim roomUncalibrated; assumes your screen follows sRGB / P3

Reference values in the literature

These are cited for context. After each run the summary includes a hedged "in theory" note that compares your score with them, as if the demo measured like the real test. It hasn't been shown to, so the note is not a prediction.

  • EASA guidance (AMC1 MED.B.075) and the UK CAA name CAD thresholds below 6 SN (deutan) or below 12 SN (protan) as acceptable for applicants who fail pseudoisochromatic plates. A yellow-blue value above 2 SN is investigated further.
  • Barbur & Rodriguez-Carmona (2017) propose grades at 2.35, 4, 6 and 12 SN.

What is reconstructed or assumed

  • 1 SN in CIE xy (≈ 0.0041 red-green, ≈ 0.0128 yellow-blue) is derived from published threshold cone contrasts, not published directly.
  • Staircase step sizes and stopping rule are assumed and were tuned only for plausible duration. That is an internal consistency check with simulated observers, not a validation with people.
  • Scores are the mean ratio of your thresholds to the reconstructed reference thresholds. The original software uses its own model.
  • Display: no calibration, unknown absolute luminance, 8-bit output extended with dithering. Results depend strongly on your screen and settings.

Privacy

The app runs entirely in your browser. Profiles, setup and runs are stored only in this browser (IndexedDB, or localStorage as a fallback) and never leave your device unless you export a backup yourself. No cookies are set.

To see whether the demo is used, anonymous usage statistics go to a self-hosted Umami instance on the same server (umamii.yzi.ch). It is cookieless and counts page views and events such as "profile created", "run started" or "run finished" with the run mode and trial count. Profile names, PINs, screen setup and scores are never sent. As with any website, the hosting server may also keep short technical logs (such as IP address and time) to deliver the files.

Disclaimer

Provided as is, for demonstration and education only. It is not a medical device, is not intended for diagnosis, screening, monitoring or prediction of any condition, and must not be used for decisions about health, fitness to fly, employment or certification.

Sources

  • City Occupational Ltd — The CAD test: what one needs to know (operator manual, 2010)
  • Barbur & Rodriguez-Carmona — Colour vision requirements in visually demanding occupations, Br Med Bull 2017
  • Waveguiding cone photoreceptors and colour vision in diabetes (stimulus geometry, timing, background)
  • Colour vision changes across lifespan: FM100 and CAD (the 16 colour directions)
  • Carter & Barbur — Colour vision assessment for maritime lookout (UK MCA review, 2015)
  • Dr Adrian Chorley — The CAD test and how it works (2016)
  • UK CAA — Colour vision guidance material