About

Palette Vault is a catalogue of four-color palettes, a set of free color tools, and a Chrome extension. Everything is free to use and nothing requires an account.

How the palettes are put together

Each one is built in OKLCH, a perceptually uniform color space, which is the reason they look considered rather than arbitrary: in OKLCH an equal step in lightness looks like an equal step to the eye, so a ramp holds together where the same approach in RGB or HSL falls apart.

Within that, three things do the work. The four hues follow a classical relationship — monochrome, analogous, complementary, split complementary, triad or tetrad — rather than sitting at arbitrary angles. Lightness is laid out as a deliberate ladder from light to dark, so every palette has somewhere to put text. And chroma tapers at both extremes, because very light and very dark colors physically hold less of it, which keeps the pale end from going chalky and the dark end from turning muddy.

Tags come from the result rather than the intent: a palette is listed under Blue because its colors measure as blue, not because anything set out to make a blue palette. There are 43 categories in all, covering both moods — pastel, vintage, neon, night — and color families.

Licence

Colors cannot be owned, and these are free for anything.Commercial work, client projects, products for sale — no permission and no attribution needed. A palette is four numbers; there is nothing here to license.

Likes and privacy

Like counts are the only thing stored on a server, in a Firebase Realtime Database that holds nothing but a number per palette. Your collection lives in your own browser's local storage and is never sent anywhere. There is no analytics script, no cookie banner, and no account to create — the site cannot identify you because it never learns anything about you.

The tools

Eight of them, all running entirely in your browser: a palette generator, an image extractor, a WCAG contrast checker, a color picker, the Tailwind palette, every named CSS color, a gradient library and a gradient builder. Images you drop into the extractor are never uploaded.

The image extractor is worth a note on method. It clusters pixels with k-means in OKLab rather than counting frequent RGB values, because distance in OKLab tracks how different two colors actually look. Cluster a landscape photograph in RGB and you get five near-identical browns while the one red jacket that defines the shot disappears.

Open source

The site and the extension are on GitHub. Corrections and issues are welcome there.