Image palette extraction — Input checklist

Worked example

Pick an image that represents the look you want, and crop away big neutral areas first because the extractor counts pixels.

The tool shrinks the image so its longest side is at most 160 pixels, ignores pixels that are mostly transparent, rounds every channel to steps of 24, counts the groups, and lists up to eight swatches by frequency. Two groups closer than about 58 units in RGB distance are merged, so near-identical shades will not fill the list.

Imagine an interior blogger who wants a site palette from a hero photo of a living room. If half the frame is a white ceiling and a grey floor, the first swatches will be white and grey. Crop to the sofa, cushions and plants, and the list will show the greens, terracottas and creams you actually care about.

The common mistake is choosing a busy photo with many unrelated objects and expecting a tidy five-colour scheme. Pick one image with a clear subject and lighting, avoid heavy Instagram-style filters unless you want those tints, and use a logo or illustration for flat, repeatable results.

A quick check before running it: squint at the image and name the three colours you expect. If you cannot, the extractor will not know either. Then compare your guess with the first swatches and see whether a tiny accent, such as a red button, is missing; if so, crop closer to that accent and run it again.

Image with several dominant colors

Palette swatches and color codes

Only opaque pixels count, so a transparent PNG gives you the artwork colours and ignores the empty background.