On-device AI · WebGPU

Bring old photos back to life.

Repair, sharpen, fill, and reframe — a full AI photo studio that runs entirely in your browser. Your photos never leave your device.

Detecting GPU…
100% private
Free & open
Why Restore

A restoration studio, minus the upload.

The models come to you. Nothing to install, nobody to trust.

Five tools, one studio

Restore, sharpen, fill, ID photos, and smart crop — every photo fix in a single page.

100% private

Old family photos are personal. Everything runs on your device — no uploads, no servers.

WebGPU accelerated

Inference runs on your GPU when available, with automatic WASM fallback everywhere else.

Real AI models

Swin2SR super-resolution, GFPGAN face revival, LaMa inpainting, BiRefNet & MODNet matting — locally, MIT/Apache licensed.

How it works

Three steps. Zero uploads.

  1. 01

    Drop in a photo

    Drag & drop, paste, or browse. JPG, PNG, or WebP — from scans to phone snapshots.

  2. 02

    Pick a mode

    Restore old photos, sharpen blur, brush away objects, swap ID backgrounds, or crop smart.

  3. 03

    Compare & download

    Slide between before and after, then save a full-quality PNG. Models cache for instant reuse.

Restore

Scratches, noise & color revived

Sharpen

Swin2SR 2×/4× super-resolution

Fill

LaMa inpainting erases objects

ID Photo

Cutouts & passport sizes

Crop

Subject-aware smart framing

Complete guide

About the photo restorer

A free AI toolkit for restoring old and damaged photographs: scratch removal, blemish repair, face enhancement, and upscaling in one place, running fully on your device. It is designed for digitized family archives, historical scans, and damaged prints where uploading sentimental images to a third party feels wrong — and usually costs a per-photo fee anyway.

How it works

The toolkit combines several on-device models: a LaMa-style inpainting network fills scratches, tears, and dust spots by reasoning about the surrounding pixels; a face-restoration model sharpens and repairs portrait detail; and super-resolution enlarges the cleaned result. Masking scratches yourself gives the inpainter precise guidance, while one-click modes handle common defects automatically. All models run via WebGPU when the browser supports it, with WebAssembly fallback.

Limits & requirements

First runs download model weights, which are cached afterward. Restoration quality depends on the scan: deeply faded or water-damaged photos benefit from manual masking. Keep your original scans — all edits produce a new file rather than overwriting the source.

Privacy

Family photos can be deeply personal. Here they never leave the browser: no upload, no server-side processing, no retention policy to worry about.

Support

Questions, answered.

No. Every model runs in your browser. Models download once from Hugging Face and are cached locally — after that, all processing is 100% on-device.