Copy text from images (OCR)

How do I get text out of an image?

Drop or paste the image above and check its language. The text appears beside it a few seconds later. Copy it, or save it as a .txt or Markdown file. The open-source Tesseract engine reads the image in your browser, so the image is never uploaded.

1. Add the picture. Drop files or a whole folder, pick them, or paste a screenshot with Ctrl V (⌘ V on a Mac).
2. Check the language. English plus your browser’s language is chosen for you. Add any of the 123 others by typing their name.
3. Copy or download. Choose lines as on the page, joined paragraphs or Markdown. The tool marks words it was unsure of, so you can check them against the picture.

How do I copy text from a screenshot?

Take the screenshot to your clipboard instead of a file, then paste it here. On a Mac, press ⌘ Ctrl Shift 4 and drag over the text. On Windows, press Win Shift S. Back on this page, press Ctrl V or ⌘ V. The tool enlarges small screenshots and flips dark-mode ones to dark-on-light before it reads them, because the engine reads both better that way.

How accurate is it?

On clean printed text it gets nearly every character right. On our own test pictures (synthetic pages, chat screenshots and a tilted, unevenly lit photo of a page) it read 98 to 100% of characters correctly. Accuracy drops with blur, very small or decorative type, and handwriting, which Tesseract is not trained for. Hover a word in the text to see where it came from in the picture. Words in yellow are the ones to check.

Which languages can it read?

It reads 123 languages and scripts, from Arabic and Bengali to Vietnamese and Yoruba. That includes Chinese, Japanese and Korean (also in vertical writing), Hindi, Hebrew and Russian. You can mix up to six in one picture. Your browser downloads each language (0.5 to 6 MB) from jsDelivr the first time you use it, then keeps it. Next time it starts at once and works without a connection.

Can it read scanned PDFs?

Yes. The tool draws each page at about 300 dpi and reads it like a picture. It reads up to 500 pages at once, and the download joins the pages into one file. If your PDF already has selectable text, PDF to Text gets it out exactly and much faster.

How does it compare with other free OCR sites?

	This tool	OnlineOCR.net (guest)	i2OCR
Where the image goes	Stays on your device	Uploaded, deleted after conversion	Uploaded
Limits	None beyond your device’s memory	15 MB per file, 5 files an hour	One image at a time on the free plan
Languages	123	46	Claims 128
Batches	Yes, with a .zip out	Registered users only	Premium only

Checked on 4 November 2026 on onlineocr.net and i2ocr.com. Both offer Word and PDF output, which this tool does not.

How do I get better results?

• Crop to the text. Borders, photos and busy backgrounds give the engine more to misread.
• Choose every language in the picture, and only those.
• For a photo of paper with shadows, turn on Even out shadows under Settings.
• For app screens, slides and signs, set the page layout to Scattered text.
• If the picture is sideways, press R to turn it and read it again.