Click anywhere on the image to sample that exact spot.
How the colors are chosen: the image is sampled and the colors grouped by similarity using
k-means clustering, which finds the shades that genuinely dominate rather than just picking the
brightest pixels. Clustering runs in LAB color space, which spaces colors the way an eye judges
difference, so the results match what you actually see rather than what the raw numbers suggest.
Raise the color count for a more detailed palette, lower it to find the core two or three.
Using them: click any swatch to copy its hex, or grab the whole set as CSS custom properties
or JSON. For design work, the largest block usually wants to be a background, the smallest and most
saturated makes a good accent, and anything in between works as supporting tone. Check contrast
before you set text on any of them — a palette pulled from a photo is not automatically readable.