HomeArticlesColours & Light

Colour Vision & Colour Blindness: The LMS Model Explained

Three overlapping cone responses, recoded into red-green and blue-yellow opponent channels — and what protanopia, deuteranopia, tritanopia and achromatopsia actually remove from that picture.

mysimulator teamUpdated June 2026≈ 6 min read▶ Open the simulation

Three cones, one trick of overlap

Human colour vision is trichromatic: the retina has three types of cone cell, each with a photopigment sensitive across a broad but different range of wavelengths, conventionally labelled by their peak sensitivity — S (short, ~420 nm, "blue"), M (medium, ~534 nm, "green") and L (long, ~564 nm, "red"). None of these curves is narrow; they overlap heavily, especially M and L, which peak only about 30 nm apart. Colour perception isn't three independent channels reporting wavelength — it's the brain comparing the relative response across all three, which is why a single wavelength and a mixture of wavelengths can look identical (metamerism) as long as they produce the same S:M:L response ratio.

live demo · LMS cone response bars reacting to a shifting spectrum● LIVE

Downstream of the cones, the visual system recodes these three signals into opponent channels: a red-green axis (roughly L minus M), a blue-yellow axis (roughly S minus L+M), and a black-white luminance channel. This opponent-process stage, not the raw cone responses, is what most colour-blindness effects and colour-contrast illusions actually operate on.

What actually goes missing in colour blindness

The common inherited forms are anomalies or absences of the L or M cone photopigment, both encoded on the X chromosome (which is why red-green colour blindness affects far more men than women). Protanopia is a complete absence of L-cone function; deuteranopia is a complete absence of M-cone function. Both collapse the red-green opponent axis, since it depends on comparing L and M, but they aren't identical — protanopia also dims perceived brightness in the red end of the spectrum because L cones normally contribute to luminance, while deuteranopia does not.

Tritanopia, absence of S-cone function, is rare and not sex-linked; it collapses the blue-yellow axis instead, so blues and yellows become confused while red-green discrimination stays largely intact. Achromatopsia, true absence or near-absence of all cone function (relying on rods alone), is very rare and eliminates colour vision almost entirely, leaving only luminance-based, low-acuity vision.

condition        cone affected     axis collapsed        typical confusion
protanopia       L absent          red-green              red ↔ dark green/brown
deuteranopia     M absent          red-green              red ↔ green (brighter than protan)
tritanopia       S absent          blue-yellow            blue ↔ green, yellow ↔ violet
achromatopsia    all cones         both                   greyscale vision only

Simulating it: LMS transform, not just a colour filter

A perceptually honest colour-blindness simulator doesn't just desaturate or hue-shift an image; it works in LMS space. First the image's RGB values are converted to an estimate of actual L, M and S cone excitation using a standard transform (such as the Hunt-Pointer-Estevez matrix). Then, for the deficiency being simulated, the missing cone's response is estimated from the other two — for example, under a common protanopia model, the L response is replaced by a linear combination of M and S. Finally the modified LMS triple is converted back to RGB for display. This reproduces genuine confusion lines — the actual sets of colours a dichromat cannot distinguish — much more faithfully than a naive hue rotation, which tends to either under- or over-state the confusion.

Why Ishihara plates work the way they do

The classic Ishihara test embeds a numeral in a field of coloured dots, where the numeral's dots and the background dots are chosen to differ mainly along the red-green confusion axis while being matched in brightness and saturation as closely as possible. A trichromat sees the number easily because their L and M signals differ. A protanope or deuteranope, missing that axis, sees the numeral and background as nearly the same perceived colour and the number vanishes into visual noise — the test is essentially a direct, deliberately low-tech probe of exactly the LMS-to-opponent-channel pathway described above.

Frequently asked questions

Why is red-green colour blindness so much more common in men?

The genes for the L and M cone photopigments sit on the X chromosome. Men have only one X chromosome, so a single defective copy causes the deficiency; women have two X chromosomes and usually carry at least one working copy, which is why protanopia and deuteranopia affect roughly 1 in 12 men but only about 1 in 200 women.

Is colour-blindness the same as seeing in greyscale?

No, except in the rare case of complete achromatopsia. Protanopia, deuteranopia and tritanopia each eliminate one specific opponent colour axis while leaving the others intact, so a deuteranope still perceives blue-yellow contrast normally — they simply lose the ability to distinguish colours that differ mainly along the red-green axis.

Can a simple hue-rotation filter accurately simulate colour blindness?

Not accurately. A faithful simulation converts the image to estimated LMS cone-response values, zeroes or reconstructs the missing cone's contribution from the remaining two, and converts back to RGB. This reproduces the real confusion lines a dichromat experiences; a naive RGB hue shift only approximates the effect and can mislabel colours that would actually be confused.

Try it live

Everything above runs in your browser — open Colour Vision Simulator and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.

▶ Open Colour Vision Simulator simulation

What did you find?

Add reproduction steps (optional)