AI Photo Colorizer
Colorize & restore black-and-white photos — on-device AI.
Runs entirely in your browser — nothing you enter is uploaded, logged, or tracked.
A real AI photo colorizer that runs 100% on your device — drop in a black-and-white photo and an on-device neural network (no servers, no upload) predicts realistic colour for the whole scene. It rebuilds colour by combining the AI’s colour estimate with your photo’s original brightness detail, so the result stays crisp (large photos are scaled to 1600 px on the long edge first). Fine-tune the saturation and warmth, drag the divider to compare against the original, switch between a natural and a more vivid model, and export a PNG, JPEG or WebP. The Natural model downloads once (~123 MB); the optional Vivid model is a separate ~130 MB download; after that it all works offline forever. Your photos never leave your device, and the export carries no GPS or camera metadata.
Export
Lossless still image.
Re-encoded on your device — no GPS or camera metadata, no watermark.
Frequently asked questions
Is my photo uploaded anywhere?
No. The AI colorization model and all processing run inside your browser on your own device — there are no servers, no upload, no analytics. After the one-time model download you can disconnect from the internet and it still works. That privacy is the whole point: cloud colorizers send your (often personal, old family) photos to their servers; this one never does.
How does an AI colorize a photo in my browser with nothing uploaded?
It uses a classic colorization neural network (Zhang “Colorful Image Colorization”) compiled to run on your device via WebAssembly. The first time you use the tool it downloads the model once (~123 MB, shown with a progress bar), caches it, and from then on every colorization happens locally and offline. The AI predicts the colour channels; your photo’s original brightness is kept for full-resolution sharpness.
How accurate are the colours?
They’re a plausible AI estimate, not the true original colours — the model has never seen your scene. It’s usually convincing on skin, sky, foliage and common objects, but ambiguous things (a shirt, a car, a painted wall) can come out muted or simply a different colour than reality. The saturation and warmth controls and the natural/vivid model toggle let you steer it.
Will it stay sharp?
Yes, within the working size. The AI predicts colour at a small internal resolution, but the tool recombines that colour with your photo’s own brightness detail, so the result is as sharp as the input — not a blurry upscale. Very large photos are first scaled down to 1600 px on the long edge (plenty for most old-photo scans and for sharing), so it’s not a tool for preserving 24-megapixel originals at full size.
Can I colorize a photo that already has some colour?
Yes — it will re-imagine the colours from scratch based on the brightness, which can be a fun way to recolour a faded or tinted photo. It’s designed for black-and-white input, but it works on any photo.
Does it remove EXIF / GPS metadata?
Always. The export is re-encoded from raw pixels, so the original’s GPS location, camera model and EXIF tags can’t carry over — there’s nothing to toggle.