Image to Text (OCR)

How to use Image to Text (OCR)

1. 1

   Add the image — a screenshot, a photo of a page, or a scan.

2. 2

   Pick the language the text is written in, so recognition is accurate.

3. 3

   Click Extract text and watch the OCR engine work through the image.

4. 4

   Copy the recognized text or download it as a .txt file.

Turn a photo, screenshot or scan into text you can edit

"Image to text" and "extract text from image" are two ways of asking the same thing: a picture contains words, and you need those words as real, selectable text — to paste into a document, search, translate or fix a typo. This tool does exactly that with OCR (optical character recognition): it looks at the pixels, works out which shapes are letters, and hands you back the text as characters.

Typical inputs are a screenshot of an error message, a photo of a printed page or whiteboard, a scanned letter, a menu, a recipe or a slide someone pasted as an image. Anything JPG, PNG, WEBP or similar works. The output appears in an editable box you can copy in one click or download as a .txt file, so the extracted text goes wherever you need it.

Getting accurate results: language, sharpness, angle

The single biggest lever is the language selector. OCR engines use a language model to resolve ambiguous shapes — is that an l, an I or a 1? — so telling it the text is Spanish or German changes the outcome noticeably. English, Spanish, French, German, Italian and Portuguese are available; pick the one the text is actually written in, not the one your browser uses.

After that, image quality. Clean screenshots and flat, well-lit scans typically recognise above 95 % of characters correctly. Photos taken at an angle, under a lamp that casts a shadow, or of glossy paper do worse. If a result is messy, retake the photo straight on with the page filling the frame, and crop away anything that is not text — a photo of a page with a coffee cup in it gives the engine less to misread without the cup.

• Screenshots and PDFs rendered as images → near-perfect recognition
• Flat scans at 200-300 DPI → excellent
• Phone photos, straight on and in focus → good
• Handwriting, decorative fonts, curved pages → partial at best

What OCR cannot do — and what to do instead

OCR reads printed characters; it is not a handwriting reader and it is not a layout engine. Cursive notes, signatures and most handwriting will come back as fragments. Tables lose their columns — the words are recognised, but you will get lines of text rather than a grid, so paste them into a spreadsheet and split them there. Text laid over a busy photo, or in very light grey on white, may be skipped entirely.

For multi-page documents, the better route is often the PDF tools: a scanned PDF can be converted to PNG page by page and each page run through OCR, or — if the PDF already has a text layer — the text can be copied without OCR at all. And if what you actually need is the picture without the text rather than the text without the picture, the image editing tools are one click away in the header.

Why your image never leaves your device

Every step runs inside your browser, using the WebAssembly and Canvas APIs your device already ships with. No file is uploaded, no copy is kept, and nothing sits in a queue on a server waiting to be deleted — which is also why the tool keeps working with the network disconnected.

OCR is a case where that guarantee is worth checking, because the images people extract text from are IDs, bank letters, contracts, medical results and private messages. Here the recognition engine is Tesseract compiled to WebAssembly: the first run downloads the engine and the chosen language model (a few megabytes), your browser caches them, and every image afterwards is processed entirely inside the tab. Disconnect from the network after that first download and the tool still works.

There is no account to create and no page counter to run down. The cost of each extraction is a few seconds of your own CPU time — more on a phone than a laptop, but the same privacy on both.

Frequently Asked Questions

Related tools