Read the words out of a photo, a screenshot or a scan and get text you can actually paste.
Image to Text (OCR)
Read the words out of a photo, a screenshot or a scan and get text you can actually paste.
The images are decoded and read inside this tab, and the text never leaves it. Only the recognition data for the language you choose is fetched, once, and your browser caches it.
The commonest reason to read a picture is that somebody sent a screenshot of something they could have sent as text. So Ctrl+V or ⌘V anywhere on this page takes whatever is on the clipboard and starts there, no saving to disk first. Up to twenty images queue up and are read one at a time, each with a confidence figure of its own, and the result comes back either exactly as the lines sat on the page or reflowed into paragraphs: words hyphenated across a line break mended, list items kept on their own lines, Chinese and Japanese joined without spaces between them. Recognition runs on your device. The first time you choose a language its recognition data downloads once — a few megabytes — and your browser keeps it for next time.
Features
Paste a screenshot with Ctrl+V or ⌘V anywhere on the page
Up to 20 images in one queue, read in turn with progress shown per image
Twelve languages, plus an optional second language for a mixed-script page
Keep the line breaks as read, or reflow into paragraphs and mend hyphenated words
An enhance pass: greyscale, automatic levels, and 2× for small screenshots
A confidence figure per image, colour-coded, so you know what to check
Copy one result, copy the lot, or save everything as a single .txt
PNG, JPEG, WebP, GIF and BMP
How to use the Image to Text (OCR)
Drop images in, or paste a screenshot straight from the clipboard
Choose the language in the picture, and a second one if the page is mixed
Decide whether to keep the line breaks or reflow into paragraphs
Read the batch, then copy the text or save it as a .txt
Example
Input
whiteboard.jpg, receipt.jpg, error-dialog.png — English, reflowed into paragraphs
Output
3 results at 91 %, 84 % and 99 % confidence — saved as image-text.txt
In the combined file each result sits under its own file name.
Common errors & troubleshooting
Nothing happened for a while on the very first image. — The recognition data for that language was being fetched, which happens once and takes a few megabytes. Later images in the same language start straight away, and so does your next visit.
Handwriting came back as gibberish. — This is built for printed and typeset text, and reads handwriting far less reliably. Handwriting to Text is tuned for notes and cursive; either way, check names and numbers by eye.
Small text in a screenshot was misread. — Leave the enhance option on — it doubles pictures under 1800 px on the long side. Better still, take the screenshot again at a larger zoom level; more pixels beat any amount of processing.
Two columns ran together into one line each. — Lines are read straight across the picture, so side-by-side columns interleave. Crop each column with Image Cropper and read them one at a time.
Frequently asked questions
Do the pictures themselves get sent away to be recognised?
No. Recognition runs inside this tab and the images stay on your device. The one thing fetched from the network is the recognition data for the language you pick, and that carries nothing of yours with it.
Why does choosing a language download a few megabytes?
Because the shapes of that script's characters have to be on your machine before anything can be read locally. It is fetched once per language, cached by your browser, and reused every time after that.
Can it read two languages on the same page?
Yes. Pick a second language and both are used together, which is what a mixed English and Hindi page or a bilingual form needs. Two is the limit — more languages at once make every result worse.
What does the confidence percentage actually measure?
It is the average certainty across the characters recognised, not a score for how correct the meaning is. High confidence on a blurry photo can still be wrong, so it is best read as a hint about which results deserve a second look.
What does reflowing into paragraphs change?
It joins the lines of each paragraph into one, mends a word split by a hyphen at a line end, leaves bullets and numbered items on their own lines, and keeps blank-line breaks between paragraphs — so the text pastes as prose rather than as a column.
Can it do a whole PDF rather than a picture?
Use PDF OCR for that: it renders every page, reads it and writes a searchable text layer back into the document. This tool takes images and gives you text to paste.
Related tools
PDF OCR — Recognise the text in a scanned PDF and make it searchable.
Handwriting to Text — Read handwritten pages into typed text, keeping each word where it was written.
PDF to Text — Extract selectable text from a PDF as plain text or Markdown.
Image Cropper — Crop an image to any area or fixed aspect ratio and download it, in your browser.
Image Optimizer — Re-encode and shrink images (JPEG/WebP/PNG), optionally resizing.
Scan to PDF — Use a camera as a document scanner and save the pages as a PDF.
Image Format Converter — Convert images between PNG, JPEG and WebP entirely in your browser.
EXIF Viewer & Remover — View and strip EXIF metadata (including GPS) from photos — nothing is uploaded.
More Images tools
Image Optimizer — Re-encode and shrink images (JPEG/WebP/PNG), optionally resizing.
Image Format Converter — Convert images between PNG, JPEG and WebP entirely in your browser.
EXIF Viewer & Remover — View and strip EXIF metadata (including GPS) from photos — nothing is uploaded.
Base64 Image Encoder / Decoder — Convert images to Base64 Data URIs with ready-made HTML and CSS snippets, or decode a Data URI back to a downloadable image.
App Icon Generator — Turn one image into every iOS, macOS and Android app icon size — downloaded as a ZIP, in your browser.