By ColorUI Team6 min read
How to Extract a Palette From an Image (and Use It Well)
A peek under the hood of k-means clustering, color moods, and how to turn any photo into a usable design system in seconds.
The algorithm
Most palette extractors use k-means clustering. The idea: pick K (say, 5) random color centers, assign every pixel to the nearest center, recompute each center as the average of its pixels, and repeat until the centers stop moving.
Color moods
A raw extraction often looks muddy. Apply a mood preset to push the palette toward a feel:
- Colorful - boost saturation.
- Bright - boost saturation and lightness.
- Muted - drop saturation.
- Deep - drop lightness, keep chroma.
- Dark - drop lightness aggressively.
Try it on your own photo at Extract from Image.