Colorize photos

The archive that never had colour to begin with

Millions of photographs exist only in black and white, not because someone chose monochrome as a style but because colour film was expensive, unreliable, or simply hadn't reached ordinary use yet at the time they were taken. Family archives, newspaper morgues, and municipal records are full of these images, and for decades the only way to add colour was tedious manual hand-tinting, one photo, one brushstroke at a time. This endpoint does the equivalent job automatically, at the pace an archive actually needs.

How the guess gets made

The task reads the grayscale image and infers colour from context — skin reads as skin tone, sky reads as sky blue, foliage reads as green — the same kind of visual reasoning a skilled colourist uses, applied consistently and quickly. You send the black-and-white photo, the task returns a coloured version, and because it's a plausible reconstruction rather than a record of the true original colours, results are best treated as an enhancement for viewing and sharing, not as historical proof of what something actually looked like.

A century of trying to solve this by hand

Hand-tinting black-and-white photographs dates back to the earliest days of photography itself, well before colour film existed, and colourization of old film and photos has remained a recognisable craft ever since — practiced by specialists for historical restoration, film remastering, and personal archives. Automating that judgment call is what turns a slow, specialist skill into something an archive of any size can apply consistently.

Where this actually gets used

Genealogy platforms use it to make century-old family portraits feel present rather than distant. Museums and local history projects use it to make archival photography more approachable to a general audience. Media and publishing use it to refresh archival photography for retrospective features without commissioning new hand-tinting work for every image.

Working it into a digitisation pipeline

The endpoint runs asynchronously: submit the scanned or digitised photo, get a task_id, and the coloured result arrives by signed webhook the moment it's ready, or through a signed link if that fits your flow better. It pairs naturally with a preceding scan or upscale step and a following export step, so a physical photo can move from scan to sharp, coloured, publish-ready image without manual handling in between.