opticquiz-cvd (Python)

Is your chart colorblind-safe? Red-green color-vision deficiency affects ~1 in 12 men — and the classic failure is a plot where the red series and the green series look identical to those readers. This checks any palette for that, fixes it if it fails, and also checks WCAG contrast — with zero dependencies, dropping straight into a matplotlib / plotly / seaborn workflow.

It simulates protanopia, deuteranopia and tritanopia (Machado 2009, or Brettel 1997) and scores perceptual difference with CIEDE2000, flagging pairs that are distinct to normal vision but collapse under a color-vision-deficiency simulation.

Method (citable): https://doi.org/10.5281/zenodo.21310578

Install

pip install opticquiz-cvd

Check a chart palette

import opticquiz_cvd as cvd

report = cvd.check_palette(["#d7191c", "#1a9641", "#2166ac"])
report["pass"]                         # False
report["types"]["deutan"]["conflicts"]
# [{'a': '#d7191c', 'b': '#1a9641', 'normal': 70.6, 'sim': 8.1, 'severity': 'risk'}]

Works directly with matplotlib colors (hex, or 0-1 / 0-255 RGB tuples):

import matplotlib.pyplot as plt, opticquiz_cvd as cvd
cycle = plt.rcParams["axes.prop_cycle"].by_key()["color"]
print("colorblind-safe:", cvd.check_palette(cycle)["pass"])

check_palette(colors, distinct=13, collapse=10, severity=1.0, model="machado") — severity (0–1) checks milder anomalous trichromacy; model is "machado" or "brettel".

Fix a failing palette

fixed = cvd.fix_palette(["#d7191c", "#1a9641"])
fixed["colors"]   # ['#c80011', '#2da24c']  — now colorblind-safe
fixed["drift"]    # [4.0, 4.1]  — how far each color moved (CIEDE2000)
fixed["pass"]     # True

Check contrast (WCAG)

cvd.check_contrast("#767676", "#ffffff")
# {'ratio': 4.54, 'AA': True, 'AAA': False, 'ui': True, 'pass': True}
cvd.contrast_ratio("#000000", "#ffffff")   # 21.0

Simulate

cvd.simulate("#d7191c", "deutan")               # '#8a7b0c'  (Machado, full)
cvd.simulate("#d7191c", "deutan", 0.5)          # milder
cvd.simulate("#d7191c", "deutan", 1.0, "brettel")
cvd.delta_e("#d7191c", "#1a9641")               # 70.6

Command line

python -m opticquiz_cvd "#d7191c" "#1a9641" "#2166ac"
# FAIL - color conflicts found (3 colors)
#   deutan: #d7191c/#1a9641 dE8.1(risk)

Honest scope

Simulates a model of color vision — an approximation of a diverse population, not any single person's vision — and results depend on an uncalibrated screen. Severity below 1 is a disclosed approximation of anomalous trichromacy. It is not a legal accessibility audit and does not certify ADA / Section 508 / WCAG / EU Accessibility Act compliance.

License

MIT. Methods: Machado, Oliveira & Fernandes (2009); Brettel, Viénot & Mollon (1997); CIEDE2000; WCAG 2.x. Part of OpticQuiz.