Extract text from an image
Turn a photo or screenshot into text you can select, search and paste. Add the image and the recognised text appears ready to copy.
A photo of a page, a screenshot, or a scan
Works on scans and photos where the text is part of the picture and cannot be selected.
When you need this
A photograph of a page, a screenshot of an error message, a receipt, a whiteboard after a meeting, a slide from a presentation you could not download. In all of these the words are visible but locked inside pixels, so you cannot search them, copy a line, or paste them anywhere. Optical character recognition reads the shapes and gives you the text back.
How to get an accurate result
Recognition quality depends far more on the photo than on the software. Shoot straight on rather than at an angle, get even light without a shadow falling across the page, and fill the frame with the text so the letters are as large as possible. A sharp phone photo of a printed page is usually read almost perfectly; a dim, angled shot of the same page can produce nonsense.
What it will struggle with
Handwriting is the hard case and should be treated as unreliable, since the models here are trained on printed type. Decorative and script fonts, text over busy photographs, very low-resolution screenshots, and columns that interleave in complex layouts will all produce mistakes. Always read the output before trusting it, particularly for numbers, where a misread digit looks perfectly plausible.
Frequently asked questions
Which languages can it read?
English, Spanish, French, German and Portuguese. Pick the language before running it, because the choice genuinely changes the result: the recogniser uses the language to resolve letters that are ambiguous in isolation, so a French page read as English will contain avoidable errors.
Can it read handwriting?
Not reliably. Neat block capitals sometimes come through, but ordinary cursive generally does not. The models are trained on printed text, and handwriting is a genuinely different problem.
Does it keep the layout of the original?
No, and that is deliberate. The output is plain text, because plain text is what people actually want when they cannot select anything in an image. Line breaks are broadly preserved but columns, tables and text boxes are flattened into reading order.