klar-plugin-cvd-machado

A klar plugin that measures whether two colors stay distinguishable for a person with anomalous trichromacy (partial color blindness — the most common form), using the Machado, Oliveira & Fernandes (2009) model at extreme severity.

It adds two contrast types:

| --type | Deficiency | |---|---| | cvd-deuteranomaly | shifted M (green) cone — extreme | | cvd-protanomaly | shifted L (red) cone — extreme |

What it measures

Each type simulates both input colors as that anomalous trichromat would see them, then returns the CIEDE2000 (ΔE₀₀) between the simulated colors:

• high → the two colors stay clearly separated (good)
• low → they collapse together / become confusable (a real accessibility problem)

Higher-is-better, integer-rounded — same polarity and rounding as klar's built-in deltaE.

Install

npm install -g klar-cli klar-plugin-cvd-machado

klar discovers it automatically via the klar-plugin-* convention.

Usage

$ klar contrast "#d40000" "#00a000" --type cvd-deuteranomaly -q
11
$ klar plugins list   # confirm the two types are registered

A common convention is to treat ΔE ≥ 11 as "still clearly distinct" — your policy to apply; the plugin returns the raw number.

About the "extreme" severity (and its caveat)

Machado's model is parameterized by a severity from 0 (normal) to 1 (dichromat). This plugin fixes severity at 0.8 — "extreme" anomalous (near-dichromatic) — because that's where the accessibility risk concentrates: mild anomalous trichromacy is close to normal vision, and full dichromacy is covered by the companion klar-plugin-cvd-brettel.

Caveat: Machado's severity is a spectral-shift parameter and only a rough proxy for how a given individual actually performs (optical-density variation means the mapping isn't monotonic). So 0.8 is labelled by what it physically is — "extreme anomalous" — not as "the average anomalous trichromat." The distribution of real-world severity is gene-driven and lumpy, not a bell curve.

Tritanomaly is intentionally excluded — it's S-cone based (different genetics, rare) and Machado models it poorly; tritan deficiency is served by the Brettel plugin.

Attribution & license

• Plugin code: MIT.
• Matrices from Machado, G. M., Oliveira, M. M., & Fernandes, L. A. F. (2009). A Physiologically-based Model for Simulation of Color Vision Deficiency. IEEE TVCG, 15(6). Matrix table, used with attribution. Applied in linear RGB (per DaltonLens).