Color Blind Test

Dot plates with the same brightness and different colors test color discrimination along three directions

This color blind test is an activity in which you read a two-digit number (10–99) hidden in a dot plate. On every plate the brightness of the digit dots and the background dots is mixed at random over the same range, and only the color differs. If you can see the digit, you read it from color difference, not from brightness.

The color difference is split into three directions: the direction of colors you cannot tell apart when the red cone is missing (protan), when the green cone is missing (deutan) and when the blue cone is missing (tritan). In each direction the test starts with a large difference and reduces it step by step to the Cambridge Colour Test expected value (the baseline). At the end you get how many levels you read on each axis, a plate ladder, and a simulation showing how the same plate looks with each type of color vision.

Not a diagnosis The result is a reference value from the screen you are using. The screen is not color-calibrated, so it can differ from a clinic result, and it does not decide whether you have a color vision deficiency or how strong it is.

Why test three directions separately?

The cone cells that see color come in three kinds: red (L), green (M) and blue (S). If one kind is missing or works differently, colors lying on certain straight lines (confusion lines) become hard to tell apart. The US National Eye Institute (NEI) says that trouble with red and green is the most common kind, that some people mix up blue and yellow, and that seeing no color at all is rare. Testing each direction on its own shows where a larger color difference is needed.

Reasons you might use it:

• You are curious about your color vision: if you have ever mixed up traffic lights, chart colors or map colors, you can get a rough idea of which direction is hard.
• Before an eye appointment: use it to organize which color combinations confuse you and since when. It does not replace an exam.
• You want to see through a color-deficient eye: the simulation on the result screen shows how the same plate looks to protan, deutan and tritan viewers, and how the digit fades as strength rises.
• You are choosing design colors: the plate ladder and simulation let you experience why signs that rely on red versus green alone are risky.

How to take the color blind test

1. Turn off night mode, blue-light filters and color inversion. Sit where there is no glare and set the screen brightness to a comfortable level.
2. On the first screen, check that you can see the number 12 on the practice plate. Press 1 and then 2 to get used to the controls; the practice plate does not count toward the result.
3. Choose the number of plates: Quick 13 (4 levels per axis + 1 check plate), Standard 20 (6 levels + 2 check plates) or Detailed 32 (10 levels + 2 check plates). The more plates, the smaller the steps between levels.
4. Press the two-digit number you see on the number pad or with keys 0–9. The next plate appears as soon as you press the second digit, and there is no time limit. If you can see only one digit, press it and then "d only" (for example "7 only"); if you can see nothing, do not guess: press "Can't see" (keyboard Space or Enter).
5. If you pressed the first digit by mistake, press "Erase"; to change the answer to the previous plate, press "Back" (both are ⌫ on the keyboard). The result appears after the last plate, and on the result screen "Back" lets you change the last answer.

The test does not tell you whether an answer is right. The plate order mixes the three axes and the check plates at random. The numbers are drawn from the 89 numbers between 10 and 99 other than the practice number 12, and within one run no number appears twice. Because the number has two digits, the chance of guessing right by pressing anything on a plate you cannot see is only about 1 in 90. "New plates" redraws the digits, the dot layout and the order.

How the plates are built

Background dots are gray with the same chromaticity as the screen's reference white (D65). Digit dots are that color shifted a set distance along a confusion line. A confusion line runs from a hypothetical point that would stimulate only one cone (the copunctal point) to the background color. This tool uses copunctal points based on Smith-Pokorny cone fundamentals, in CIE 1931 xy coordinates: protan (0.7465, 0.2535), deutan (1.40, −0.40), tritan (0.1748, 0). The protan and deutan values match those compiled by Moreira et al. (2018).

The brightness (luminance Y) of each dot is picked at random from six levels between 0.12 and 0.30, and it is applied equally to digit dots and background dots. The Cambridge Colour Test scatters brightness in the same way (luminance noise) so that responses rely only on the color signal (Regan et al. 1994). Dot positions and sizes are set independently of the digit, so the digit cannot show through the way dots cluster.

Color difference is measured as the distance in CIE 1976 u′v′ chromaticity, in the same unit as the Cambridge Colour Test, ×10⁻⁴ u′v′. The largest level was chosen within what an sRGB screen can show, and the smallest level (the baseline) is the Trivector expected value for that test: 100 for protan and deutan, 150 for tritan (Hasrod & Rubin 2019). The levels in between are spaced at equal ratios.

Quick divides the same range into 4 levels (protan and deutan 1,000·464·215·100) and Detailed into 10 levels (1,000·774·599·464·359·278·215·167·129·100). The side of the confusion line to which the digit color is moved is chosen at random on each plate, but large levels that would leave the screen's range are shifted only toward the inside. The 1–2 check plates have only the digit dots in bright gray, so anyone can read them regardless of color.

How the result is read

For each axis the test counts n, the number of plates on which you got both digits right. A plate with only one digit right is marked "one digit right" (½) in the plate ladder and the response table, but counts as missed in the axis estimate. It does not matter which levels you got right. Each level has only one plate, so an easy plate can be missed by accident and a hard one guessed by luck; counting plates lets the two cancel out.

• If n equals the number of levels: you read all the way to the baseline, and the multiple is set to 1.
• If n is 0: you could not read even the largest level. It is shown as "above ×10" (above ×8 for tritan).
• Anything in between: the geometric mean √(level n × level n+1) of the nth and (n+1)th levels is taken as the difference you can read, and its ratio to the baseline is shown to one decimal place.

The rules for choosing the sentence from the multiples are below. The values 1.8 and 1.5 were set by this tool so that missing a single level is not flagged separately; they are not medical criteria. If you miss even one check plate, that notice is shown before the axis result.

In the example above, red–green is 5.0, which is at least 1.8 times blue–yellow (1.9), so the red–green sentence appears, and because deutan (5.0) is at least 1.5 times protan (1.3), the deutan axis is named. In the plate ladder on the result screen, the left side is the large difference and the rightmost cell is the baseline; each cell is split into that plate's background color and digit color, so you can compare the color difference of the plates you missed.

Types of color vision deficiency and the simulation

When one kind of cone is entirely missing, it is called dichromacy (color blindness); when the cone is present but its sensitivity has shifted, it is called anomalous trichromacy (color weakness). The NEI says red–green deficiency affects about 1 in 12 men. A worldwide review puts it at about 8% of men and 0.4% of women of European descent, and 4–6.5% of Chinese and Japanese men (Birch 2012). A 2019 survey of 2,686 South Korean adults aged 19–49 found 3.9% overall (men 6.5%, women 1.1%), with deutan (2.5%) more common than protan (0.4%) (Kim & Ng 2019).

The simulation on the result screen redraws the same plate with the Machado et al. (2009) model for protan and deutan and with the Brettel et al. (1997) method for tritan. Strength 100% is dichromacy, where that cone is missing; for protan and deutan the tool interpolates the authors' table, given in steps of 0.1, to show milder deficiencies. Tritan strengths below 100% are an approximation made by blending with the original color. Because the plates are built from colors on the confusion lines, in the dichromat simulation of the same axis the digit dots and background dots become almost the same color. In the protan view the digit dots may stay slightly darker and in the deutan view slightly brighter, but that difference blends into the luminance noise of the whole plate, so the digit becomes hard to make out.

When the screen changes the result

• Night mode and blue-light filters: these reduce blue light and turn white yellowish, so the color difference on the blue–yellow (tritan) axis shrinks first. If only that axis was hard, turn the filter off and try again.
• Color correction and accessibility filters: color inversion, color filters and color-vision correction change the plate colors themselves. If you need one, do not turn it off, but remember that the result is for that condition.
• Brightness and viewing angle: a screen that is too dark or glaring, or viewed from an angle, can make colors look different. Use full screen if you want to see the plate larger.
• Device differences: the same plate looks different on each screen. This tool does not measure or calibrate your screen, and the level values assume a standard sRGB display. Because of 8-bit color rounding, the smallest level can come out up to about 8% larger or smaller than the target.

How this differs from a clinical color vision test

The UK National Health Service (NHS) advises that even if online tests exist, you should have your eyes checked by an optician if you are worried about your vision. At an appointment it helps more to describe which colors you mix up, since when, and what difficulties it causes in daily life than to quote this tool's multiples.

Common mistakes

• Assuming that reading everything means normal color vision: reading down to the baseline means you told those differences apart on this screen. The screen is not calibrated and the tool has no clinical validation, so it does not confirm normal color vision.
• Making too much of a one-level gap: there is only one plate per level, so missing one is common. On the 20-plate test, missing only the last level gives ×1.3, and the result sentence does not change.
• Comparing with another device: phones and monitors show color differently, so the same person can get different multiples. Compare only on the same device, with the same settings and number of plates.
• Worrying about a blue–yellow result with a filter on: with night mode on, the tritan axis gets hard first. Read the result together with one taken with the filter off.
• Measuring the effect of corrective glasses: glasses or lenses that change color also change the plate colors. Do not use this tool to judge a product's effect.
• Giving it to a child who does not know two-digit numbers yet: not being able to read a number is different from not being able to tell colors apart. If you are concerned about a child's color vision, ask an eye clinic about an age-appropriate test.

Changes that call for an eye exam

According to the National Eye Institute, most people with color vision deficiency are born with it, but it can also develop later from damage to the retina, optic nerve or brain, from some diseases and medicines, and from radiation treatment, and cataracts can dull color vision with age. If any of the following applies, see an eye doctor regardless of this tool's result.

• Colors you used to tell apart have recently started to look alike.
• Colors look different after an injury to the eye or head, or after you start a new medicine. Do not stop a medicine on your own; talk to the person who prescribed it.
• A child keeps having trouble learning color names or sorting by color, and someone in the family has a color vision deficiency.

Limits and cautions

• The screen is not measured or calibrated. Level values and the baseline are for a standard sRGB display, and the color difference you actually see may differ.
• The baselines 100, 100 and 150 are Trivector expected values from the Cambridge Colour Test on calibrated equipment. This tool differs in digit size, plate shape and dot layout, so the same criterion cannot be applied directly, and its own normal range has not been established by research.
• Each level has one plate and the estimate comes from the number you read. It is coarser than tests that narrow in on a threshold with repeated measurements.
• It is a task of reading two-digit numbers and pressing them, so not knowing numbers or having difficulty with the controls changes the result. Because it is a picture-viewing task, no non-visual equivalent is provided.
• The simulation is an approximation drawn from research models, and what a person actually sees varies.
• Your answers are processed only inside this page. They are not sent to a server or kept in browser storage or the address, and they disappear when you press "Start over" or close the page.

As of The research figures in this article were checked on 2026-10-20 against each paper's abstract or full text and the authors' pages.

Sources

• 01Regan BC, Reffin JP, Mollon JD. Luminance noise and the rapid determination of discrimination ellipses in colour deficiency. Vision Research 1994;34(10):1279–1299The principle of the Cambridge Colour Test: luminance noise makes responses rely on the color signal only
• 02Hasrod N, Rubin A. The Cambridge Colour Test: Reliability of discrimination trivectors in colour space. African Vision and Eye Health 2019;78(1)Trivector expected values: protan and deutan 100, tritan 150 (×10⁻⁴ u′v′)
• 03Moreira H, Álvaro L, Melnikova A, Lillo J. Colorimetry and Dichromatic Vision. IntechOpen 2018Dichromat confusion copunctal points: protan (0.747, 0.253), deutan (1.40, −0.40), tritan (0.171, 0)
• 04Machado GM, Oliveira MM, Fernandes LAF. A Physiologically-based Model for Simulation of Color Vision Deficiency. IEEE TVCG 2009;15(6):1291–1298Protan and deutan simulation matrices (table in 0.1 strength steps)
• 05DaltonLens. Understanding LMS-based Color Blindness Simulations (2021)Smith-Pokorny XYZ to LMS matrix; the 485 and 660 nm reference wavelengths for tritan in Brettel et al. 1997
• 06Kim H, Ng JS. Prevalence of Color Vision Deficiency in an Adult Population in South Korea. Optometry and Vision Science 2019;96(11):866–8732,686 adults aged 19–49 in the 2013 Korea National Health and Nutrition Examination Survey: 3.9% (men 6.5%, women 1.1%), deutan 2.5%, protan 0.4%
• 07Birch J. Worldwide prevalence of red-green color deficiency. JOSA A 2012;29(3):313–320European-descent men about 8%, women about 0.4%; Chinese and Japanese men 4–6.5%
• 08National Eye Institute (NEI). Color BlindnessCommon types, rare monochromacy, inherited and later-onset causes, cataracts, about 1 in 12 men
• 09National Eye Institute (NEI). Types of Color Vision DeficiencyColors confused in protan, deutan and tritan weakness and blindness; monochromacy
• 10National Eye Institute (NEI). Testing for Color Vision DeficiencyColor plate test, anomaloscope, color arrangement test
• 11National Health Service (NHS). Colour vision deficiency (colour blindness)Even where online tests exist, see an optician if you are worried

Frequently asked questions

Tools used together