Scan document

It turns a phone photo of a page into a flat, straight scan. AI finds the page, removes what is around it, and straightens tilt and curves, such as a curled sheet or the page of an open book. The scan keeps the page's own proportions.

Both run the same AI model through the same steps, so the scan is the same. Our server: the photo is uploaded and the scan comes back in a few seconds; you get free runs every day, then each run costs credits. Your device: nothing is uploaded and it is free with no limit; the first run downloads a 16 MB model, and the time depends on your device. If our server cannot answer, the page is scanned on your device, the tool tells you, and the run is not counted.

On your device it is free with no limit. On our server you get free runs every day, with or without an account; after that, each run costs credits, which needs an account. See the pricing page for the current numbers. Free runs reset every day at 00:00 UTC.

No. It changes the shape of the page, not its colors: shadows, stains and the color of the paper stay, and blurry or faded text stays as it is. For the best scan, photograph the whole page on a darker surface, in even light, with the page filling most of the frame.

Yes. Get the text opens the scan in Image to text. The scan is passed on inside your browser; this step uploads nothing.

No. On our server, your photo is sent over HTTPS, processed in memory, and never written to disk or stored. It is not used to train AI, and no person sees it. On your device, the photo does not leave it.

JPEG, PNG and WebP, up to 25 MB, one page per photo. You get a JPEG named after your file with -scan at the end. Its longest side is the photo's, up to 4000 px.