Free color tools · No signup · Everything runs in your browser
Color Pick
Explore the Playground
Original Color Benchmarks

Public color evidence you can inspect and reproduce

Color Pick publishes frozen questions, algorithms, datasets, results, limitations, and local runners—not opaque scores without a reference.

Last updated: July 22, 2026

Run your paletteRead the scale study

Public Palette Benchmark Platform v1.0

Direct answer: You can now run a two-to-sixteen-color palette against separate accessibility and dark-surface suites, then see exact rank and midrank percentile only within the matching color-count cohort of the frozen 450-palette corpus.

Every report names its corpus, protocol, metrics, runner, and rubric. Results remain local, downloadable as JSON or CSV, and explicitly do not create a universal palette score or accessibility certification.

Open the Public Benchmark Platform to run all suites in a cancellable Web Worker.

Tailwind Scale Accessibility Benchmark v1.0

Direct answer: OKLCH interpolation had the lowest mean adjacent-step coefficient of variation in the frozen 100-seed reference corpus under protocol v1.0. The result is specific to these weights, metrics, and sRGB output rules—not evidence that one method is universally best.

100Frozen reference seeds
400Primary scales generated
4,400Primary shade outputs
4/4Reference/holdout rank agreement

The study compares HSL, gamma-encoded sRGB, linear-light sRGB, and OKLCH under one fixed 50–950 scale protocol. It reports uniformity, hue drift, gamut mapping, and explicit white/black contrast evidence separately.

Research principles

  • Frozen versions: a published corpus or algorithm is never silently overwritten.
  • Visible evidence: direct findings, summary tables, worst cases, and limitations are readable without running JavaScript.
  • Reproducible outputs: raw JSON, CSV, manifests, checksums, engine source, and generator source are downloadable.
  • Scoped claims: every result names its corpus, sample size, protocol, metric, output gamut, and date.
  • Browser-first privacy: local runs do not require an account, database, LLM, or uploaded palette.

Open research resources

Research reportDirect answer, charts, complete method table, worst cases, limitations, and citation.MethodologyFrozen shade weights, algorithms, measurement definitions, and interpretation boundaries.Datasets and codeVersioned JSON, CSV, protocol, checksums, citation file, engine, and generator.Public Benchmark PlatformRun palette accessibility, dark-mode, cohort, and interpolation method suites locally in cancellable Web Workers.Benchmark changelogRelease history, corrections policy, compatibility, and future-version rules.Research HubCPS specifications, accessibility research, and all Color Pick evidence resources.

Versioned expansion after the first study

v3.0 adds dark-surface evidence and exact contextual cohorts without rewriting the frozen Tailwind v1.0 results. Future studies will receive separate research questions, datasets, algorithms, limitations, and versions.

No suite is folded into a universal palette score. Algorithm outcomes remain trade-offs tied to their declared protocol and intended use.

Keep it colorful

Enjoying Color Pick?

Color Pick stays free, private, and account-free. Optional donations help cover hosting, testing, and new color tools.

Donate via SociabuzzOptional. Every tool remains free.
Help improve Color Pick

Found a broken button, have a palette idea, or want a new tool? Send a quick note.

0/1200
Your message stays on your device until you choose to send it through your email app.