Colour palette from image

No. The image is read and analysed by your own browser, in a background worker, and nothing is sent to a server. The page keeps working if you go offline after it has loaded.

The image is shrunk to a small sample, then split into groups of similar colours using the median cut method in the Lab colour space, which follows how people see differences. Each group becomes one colour, groups that look almost the same are merged, and specks under 1% of the image, such as blurred edges, are folded into the nearest colour.

Flat images such as logos have only a few real colours. Instead of inventing near-duplicates, the tool merges colours that are hard to tell apart, so you may see 3 colours when you asked for 6.

Fully or mostly transparent pixels are ignored, so a logo on a see-through background gives the colours of the logo, not of the empty space. Shares are worked out from the visible pixels only.

Click or tap the image. A magnifier shows the pixels around your pointer, and the one in the middle is the one you get. With the keyboard, focus the image, use the arrow keys (Shift for bigger steps) and press Enter. Chrome and Edge also offer a button to pick any colour on your screen.

It is the same colour in eleven strengths, from 50 (very light) to 950 (very dark), with the hue kept steady. Your own colour is kept exactly at the step closest to its lightness, and the other steps are built around it. Interfaces use scales for backgrounds, borders, buttons and text.

CSS custom properties, a Tailwind @theme block, SCSS variables, JSON, and plain text, each with or without the tonal scales, plus a PNG strip of the colours. You can copy the code or download it as a file.

PNG, JPG, WebP, GIF, AVIF and BMP, up to 25 MB. Photos taken on a phone are turned the right way up automatically. SVG files are not supported here.