All tools

Image to Text (OCR)

Pull the text out of an image — drop a photo or screenshot and get copyable text, free and private in your browser.

Report a problem

Drop an image here, paste it (Ctrl/Cmd+V), or tap to choose.

or tap to choose files

Pick the language your image is written in for the best accuracy.

First read downloads the offline OCR engine and your chosen language’s data once, from Softofu’s own server, with a progress bar. After that it’s cached and works offline — each language downloads once. Accuracy depends on how clear the image is.

How to use

  1. Drop an image of some text here, paste it with Ctrl/Cmd+V, or tap to choose — you can add several at once.
  2. Choose the language your image is written in, then press “Read text” and wait while the text is recognised; the first run also downloads the engine once.
  3. Copy the text from any card, or use “Copy all” / “Download all (.txt)” when you’ve read several images.

FAQ

Are my images uploaded anywhere?

No. Everything happens right inside your browser using a small OCR engine that we host on Softofu’s own server (not a third-party CDN), so it stays reliable and can’t be taken down by anyone else. Your images are never uploaded and never leave your device. The first time you read an image, the engine and the English language data (about 6 MB together) download once — with a progress bar — then they are cached, so later reads are quicker and work even offline.

Why isn’t the text perfect?

OCR (optical character recognition) reads the shapes of letters, so the result is only as good as the picture. Clear, straight, high-contrast text on a plain background reads best. Blurry photos, tiny or stylised fonts, handwriting, busy backgrounds and skewed angles can all cause mistakes — so it’s always worth a quick proofread. Cropping tightly to just the text and using a sharp, well-lit image helps a lot.

Which languages does it support?

It reads English, Spanish, French, German, Portuguese, Italian, Dutch, Russian, Chinese (Simplified) and Japanese. Pick your image’s language from the menu before reading — that tells the engine which letters and characters to expect, which makes a big difference to accuracy. Each language downloads its own small data file once (from our server, with a progress bar) and is then cached for offline use. The result is plain text you can copy or download as a .txt file, and you can read many images in one batch — each gets its own card, with bulk buttons to combine them.