Color palette from image

How extraction works

GenHub samples your image on a canvas and runs median-cut quantization to find the most representative colors. It's the same family of algorithm used by classic palette tools, and it runs locally so your image never leaves your device.

From photo to usable palette

After extracting, lock the colors you like, reroll the rest, fine-tune any hex, then copy values or export as PNG, CSS variables or JSON. Share the result with a single link.

What median-cut actually does

Median-cut works by repeatedly splitting the image's pixels into buckets along whichever color channel has the widest spread, until it has as many buckets as colors you asked for, then averages each bucket into one representative color. It tends to find genuinely distinct, representative tones rather than just the most frequent single pixel value — which matters for photos where a huge sky or background would otherwise dominate a naive "most common pixel" approach.

Getting a better extraction

• Use a reasonably high-resolution source — extracting from a tiny thumbnail gives less accurate results.
• If the extracted palette feels dominated by one area, like a plain background, crop the image tighter to the subject first.
• Lock the colors you want to keep, then use the harmony tools to build variations around them.

After extraction

Once you have a palette, everything else works the same as building one from scratch: lock favorites, reroll the rest, edit any hex directly, check WCAG contrast against white and black, and export as PNG, CSS custom properties or JSON — or save it. Free accounts get 3 saved palettes with a public share page each; Pro removes that limit.

Where this actually helps

• Matching a brand to a product photo — pull the exact tones from packaging or a hero image instead of eyeballing them.
• Building a site theme from a mood board — extract from a reference image, then adjust for contrast before using it as a real UI palette.
• Recreating a palette from a screenshot — grab colors from a design you like without access to the original file.

A note on extracted vs. designed palettes

A palette pulled straight from a photo is a snapshot of what happened to be in that image — it's not automatically balanced for use as a UI or brand palette the way a deliberately designed one is. Treat extraction as a starting point: lock the one or two colors that feel essential, then use the harmony tools or manual hex edits to round out a palette that actually works together for your specific use.

FAQ

Related