Image to Text
Copy the text from photos, screenshots and scanned PDFs instead of retyping it.
Drag a photo, screenshot or scanned PDF here, or click to chooseTap to choose a photo, screenshot or scanned PDF
Paste a screenshot with Ctrl/⌘+V · Korean, English, JapaneseExtract text (OCR)
Recognition models download on first use. Images up to 30MB; up to 20 PDF pages per run. Check the result for recognition errors.
Accurate mode downloads a larger model. Check similar characters and URLs against the original. Selectable PDF text is extracted directly.
Open source and model licenses
How to Use
Drag in a photo or screenshot, or paste a screenshot with Ctrl/⌘+V. Scanned PDFs work too. Recognition starts right away.
Pick the language that matches the document and press Extract text again if needed. Accurate mode helps with small print.
Compare the result with the original, fix anything, then copy it or save as TXT. PDFs get a divider per page.
FAQ
How accurate is it?
Clear printed text and screenshots are mostly read correctly. Handwriting, blurry or skewed photos and decorative fonts are error-prone. Always double-check look-alike characters, URLs and numbers against the original.
How many PDF pages can it handle?
Up to 20 pages per run. For longer documents, change the first and last page and run it in parts. PDFs with selectable text are extracted directly without OCR, which is faster and exact.
Why is the first run slow?
The recognition model (language data) downloads on first use; Accurate mode uses a bigger one. It is kept in your browser, so later runs start immediately.
Are tables and formatting kept?
No. Only the text is extracted, in reading order. Table cells, fonts and bold are not preserved.
