Extract a colour palette from any image.
Drop a photo and get its six dominant colours, picked by a k-means clustering pass over the image. Click any swatch to copy the hex code, or download a labelled palette PNG with the swatches sized by how much of the image they cover.
Extract a paletteCopy hex
How it works.
Three steps, no setup, nothing uploaded.
Drop any image
A photo, a screenshot, a mood board, album art. Anything with colours worth stealing.
The clustering finds the palette
A k-means pass (with k-means++ seeding) groups every pixel into clusters and converges on the six most representative colours, weighted by how much of the image they cover.
Copy hex codes or download the swatch
Click any colour to copy its hex code to the clipboard, or download the whole palette as a labelled PNG with swatches sized by their share of the image.
Questions.
How are the colours chosen?
A k-means clustering algorithm runs over a downsampled copy of the image and converges on the six most representative colours. It finds the colours the image is actually made of, not just the brightest ones.
Why six colours?
Six is the sweet spot: enough to capture a scheme (dominant, secondary, accents) without descending into near-duplicate noise. The swatch sizes tell you each colour’s real weight in the image.
Can I use this for design work?
That is the main use: pull a palette from a mood board, a film still, a competitor’s screenshot, or a photo you love, then paste the hex codes straight into Figma, CSS, or a brand doc.
What formats do I get?
Each colour as a hex code you can click to copy, and a downloadable swatch PNG with the values labelled. Handy for sharing a palette without sending a link.
Is the image uploaded to analyse it?
No. The clustering runs in your browser on your device. Nothing is sent anywhere. Drop something confidential without a second thought.