Building a Colour Palette From a Photo You Already Love
The best palette for a project is often already sitting in an image - a photograph, a book cover, a piece of packaging that feels right. What you need is the hex codes. The Image Colour Palette Extractor reads them out of the picture.
How colours are pulled from an image
A photograph contains thousands of distinct colours, almost all of them near-duplicates. Extraction samples across the image and groups similar shades together, so the handful that actually dominate rise to the top while the noise between them is folded in.
What comes back is therefore the image's character rather than a list of pixels: the sky, the sand and the one bright object, not four hundred shades of blue.
Choosing an image that gives a usable palette
Pictures with a clear subject and a limited range work best. A sunset, a single flower, a product on a plain background, a painting with a strong mood. Busy scenes - a street, a crowd, a cluttered desk - return muddled mid-tones, because that is honestly what is in them.
It is also worth remembering that a palette from a photograph is usually all mid-lightness. Beautiful, and short of the very dark tone you need for text and the very light one you need for backgrounds. Take the colour you like best and build the full range around it with the Palette Generator.
How to extract one
Open the tool and choose an image. It is analysed on your own device, and the dominant colours come back ready to copy.
To put names to what you get - useful when writing the palette into a brief - the Colour Name Finder matches each one to its closest common name.
Before you use it
Check the combinations you intend to put together with the Contrast Checker. Colours that sit happily side by side in a photograph are not automatically readable as text on background.
Sampling a specific part instead
Extraction looks at the whole image, which is right when you want the mood of the picture and wrong when you want one particular colour from it - the exact shade of a product, a logo on a photographed sign, a single flower in a busy frame.
For that, crop the image to the part you care about before extracting. A tight crop of the subject returns the colours of the subject rather than an average of everything around it, and it is a far more reliable way of getting a specific value than hoping it survives into the top five.
Lighting is the other thing to keep in mind. A photograph records the colour under whatever light it was taken in, so the same object shot indoors and outdoors gives different hex codes, and neither is wrong.
Common questions
How many colours will I get back?
A small set of the most dominant shades - enough to characterise the image without listing every variation.
Can I extract from a screenshot?
Yes, and it is a good way to pick up the colours of a website or an app you want to reference.
Does the image leave my device?
The analysis happens in your browser. Nothing is sent anywhere.
Turn an image into a palette: