Lower coefficient of variation means the ten adjacent Delta E OK distances are more even. It is not an overall quality score.
Public Benchmark Platform · Local runners
Test a palette. Compare methods. Keep the evidence.
Run the accessibility, dark-mode, cohort, and interpolation suites in Web Workers. No color, result, or local file is uploaded or added to a public corpus.
Loading the palette benchmark platform…The versioned worker and corpus load only when requested.
Method Comparison Explorer
The second runner compares HSL, gamma-encoded sRGB, linear-light sRGB, and OKLCH under one frozen 11-shade protocol. Step variation, hue drift, gamut mapping, and explicit contrast evidence remain separate trade-offs. Color Pick does not name one universal best algorithm.
Loading the local benchmark runner…The versioned engine and Web Worker run only when requested.
How to interpret a local run
Measured circular OKLCH hue movement from the seed. Near-neutral shades below the chroma threshold are excluded.
OKLCH candidates that needed chroma reduction before final sRGB output. Mapping is evidence of a gamut trade-off.
Explicit white/black pair measurements. They are not a complete component or accessibility audit.
For a citable comparison, use the frozen 100-seed reference corpus and retain the exported manifest. A single custom seed is useful design evidence but does not become part of the published findings.