Choose. Read. Reuse.
Drop an image, select Extract Text, then check the result against your original. Copy the words into another document or save a text file.
Image tool
Image to Text
Extract the words from a photo, scan, or screenshot. Review, edit, and keep the text—all on your device.
On-device OCR
EnglishTurn pixels into editable text
Your image is processed on your device. OCR language data may be downloaded when needed, but your image is not uploaded.
You can also focus this workspace and paste a copied image with Ctrl+V or ⌘V.
0 words · 0 characters
Line and paragraph breaks are kept where OCR detects them. The result is plain text, not a replica of the image layout.
One JPG, PNG, or WebP at a time · 20 MB maximum · 24 megapixels · 10,000 px per side. No account or paid API.
From screenshots to scanned pages
Turn a JPG photo or PNG screenshot into editable English text. Clean, upright images work best. Your original file stays unchanged.
Recognized, not rewritten
No dictionary silently changes your text. OCR may misread letters and numbers, so review important content before using it.
Helpful answers
Image to Text FAQs
What image formats are supported?
Choose a JPG/JPEG, PNG, or WebP image up to 20 MB, 24 megapixels, and 10,000 pixels per side. Files are checked before OCR. Other formats must be exported to a supported format using an application that can read them.
Are my images uploaded?
No. Image contents stay on your device and are processed in your browser. The OCR worker and core load from ToolNest; English language and optional orientation model data may download from Tesseract's CDN. There is no external OCR API.
Can I extract text from screenshots?
Yes. Select or drop a screenshot, or paste a copied image while the workspace is focused. Clear, high-resolution English text gives the most useful results.
Does it work with scanned documents?
Yes, for scans saved as JPG, PNG, or WebP. It preserves recognized lines and paragraphs as editable plain text. Use PDF to Word for PDF files. Tables, multiple columns, handwriting, and unusual typography may not retain their layout.
Why does OCR sometimes make mistakes?
Blur, tiny letters, shadows, skew, handwriting, and decorative fonts can confuse OCR. Check important wording, names, and numbers. The tool does not silently correct spelling or promise an accuracy score.
Does OCR require internet?
First use needs internet to load the engine and English model data. Your browser may cache language data for later use, but offline operation is not guaranteed. Your image is never included in those downloads.
Can I copy or download the result?
Yes. Edit the recognized text, copy it to your clipboard, or download a UTF-8 .txt file. Word and character counts update as you edit. Reset removes the current image and result from the workspace.
Which languages and rotations are supported?
English is supported. The browser applies supported image orientation metadata; you can also rotate by 90, 180, or 270 degrees. Automatic orientation compares alternatives when the first reading is weak, but cannot reliably correct perspective or skew.