Color Palette from Image
Extract 5–8 dominant colors from any image using Median Cut. All processing happens locally — nothing is uploaded. Click any swatch for HEX, RGB, CMYK (sRGB), Lab D50, OKLCH and WCAG contrast values. Edit colors inline, name each swatch, and reset to the original extraction at any time. Export as ASE, CSS, JSON or SVG with optional CMYK output.
Color Palette from Image
How dominant color extraction works
Color extraction works by converting every pixel in the image to a perceptually uniform color space (Lab or OKLCH), then grouping similar colors using a clustering algorithm — typically k-means or median cut. Each cluster represents a region of the color space that appears frequently in the image. The centroid of each cluster becomes a palette color. The result is not an average of all pixels (which would produce a muddy grey-brown for most photographs) but a representative sample of the most visually significant colors. The number of clusters determines how many palette colors are extracted.
Using extracted palettes in design systems
Extracted palettes are a starting point, not a final design system. After extraction, edit individual swatches directly in the detail panel by clicking the color block — a native color picker opens. Give each color a name before exporting: named colors appear as the swatch name in ASE, as the comment in CSS, and as the role field in JSON — saving a renaming step in Illustrator or your token file. You will still typically need to adjust lightness and saturation for WCAG contrast ratios, generate tint-and-shade scales (use the Tint & Shade Generator), and verify the palette in both light and dark contexts. If you have edited swatches and want to start over, the Reset button restores the original extracted colors.
Privacy: why local processing matters
Many image processing tools upload your files to a remote server for analysis. This tool runs entirely in your browser using the Canvas API and JavaScript — your image is never transmitted. This matters for client work, unreleased product photography, medical imagery or any content that cannot leave your device. The tool accesses your image only within the sandboxed browser context and discards it when you close the tab.