AI Photo Colorizer - Colorize Black and White Photos
What colourisation is actually doing
A greyscale photograph records brightness but not hue, so the colour information is genuinely gone rather than hidden. What a colourisation model does is infer plausible colour from context. Trained across an enormous number of colour photographs, it has learned that skies above a horizon are usually blue, that foliage in that kind of texture is usually green, that skin falls in a narrow band of tones, and that a particular weave of fabric is more often one colour than another.
The result is a believable reconstruction rather than a recovery of the literal original. That is worth understanding because it explains what the tool is brilliant at and where it is merely convincing. Landscapes, portraits, street scenes and anything with familiar natural subjects come out remarkably well, because the model has seen millions of comparable examples. Distinctive man-made colours are a different matter: it has no way of knowing that a particular dress was emerald rather than navy, so it picks something plausible.
For old family archives this works beautifully. A 1940s portrait or a mid-century street photograph gains an immediacy that greyscale flattens, and the effect on a scanned family album is genuinely striking. Scan at the highest resolution you can and dust the print first, since the model reads texture as information and will happily colour a scratch.
Everything runs in your browser, which matters for exactly this use case: family photographs and personal archives are precisely the material most people would rather not hand to a third-party server. There is no account, no queue, no per-image credit, and no upload.
How to colourise a black and white photo
- Upload the greyscale image. Use the best scan you have. More detail gives the model more context to work from.
- Click Apply AI Colorization. The model analyses tones and textures across the whole frame.
- Give it a moment. Processing runs on your own hardware, so a large scan takes a little longer than a small one.
- Review and download. Compare against the original and save the result.
Two things reliably improve the outcome. Clean up dust and scratches first, since the model interprets them as image content, and even out a faded or yellowed scan with the filters tool before colourising, because a flat greyscale starting point gives clearer tonal information than a stained one. If the scan is small, run it through the AI upscaler afterwards.
Frequently asked questions
Does it recover the original colours?
It reconstructs plausible ones. A greyscale photo records brightness only, so the colour information is genuinely absent. The model infers it from context, which is why the result is convincing rather than literally accurate.
Which photos colourise best?
Landscapes, portraits and street scenes, anything with familiar natural subjects, because the model has seen enormous numbers of comparable examples. Skies, foliage and skin tones are its strongest areas.
How do I get the best result from an old print?
Scan at the highest resolution you can and clean dust off the print first, since the model reads texture as content. Evening out a yellowed scan with the filters tool beforehand also helps.
Is there a limit on how many photos I can process?
No. The model runs on your own hardware rather than a rented GPU, so there is nothing to meter. Whole family archives are fine.
Are my photos uploaded?
No. Colourisation happens entirely in your browser, which matters here because family photographs are exactly the material most people would rather not send to a third-party server.
🔎 Verify it yourself: nothing uploads
Don’t take our word for it. Open your browser’s DevTools (F12 → Network tab), then run this tool on any image. The only download you’ll see is the open-source AI model coming to your browser — your image itself is never sent anywhere. Every tool on this site works the same way — all processing happens on your device.