Image Tools
Extract Text from Image (OCR)
Reads printed text out of a photo or scan. First use downloads a small language model — after that it works instantly, even offline.
Drop an image or tap to browse
Works best on clear, printed text
Files are processed on your device. Nothing is uploaded to a server.
OCR — Image to Text: what this tool does
Extract machine-readable text from an image or scanned document. OCR is useful when the words you can see are trapped inside pixels and cannot be selected or copied normally.
How it works
- 01
Upload a photo or scan with printed text — a straight-on, well-lit shot gives the most accurate results.
- 02
We run on-device OCR to read the characters (first use downloads a small language file, cached for next time).
- 03
Copy the extracted text directly, or download it as a .txt file for use anywhere.
When this tool is useful
- Copying text from a scanned document.
- Turning a screenshot into editable text.
- Extracting notes, labels, or printed information from a photo.
Supported formats
What to know before you use this tool
OCR, or optical character recognition, turns visible characters in an image into machine-readable text. It is most useful when the source is a scan or photograph where normal text selection is not possible.
Good use cases
- Scanned documents
- Screenshots containing printed text
- Extracting text from receipts, notes, and labels
A sensible workflow
- Use a sharp, straight and well-lit source.
- Run OCR and treat the output as a first draft rather than guaranteed transcription.
- Check names, numbers, dates, URLs, and other important values against the original image.
Things worth checking
- OCR accuracy depends on image quality, language support, font, contrast, and layout.
- Handwriting and decorative typefaces are harder than clean printed text.
- If a PDF already has selectable text, OCR may be unnecessary.
Continue learning
Tips for better results
- Use a sharp, well-lit image with high contrast.
- Straighten rotated pages before OCR when possible.
- Review names, numbers, punctuation, and unusual fonts after extraction.
What to expect
OCR is not perfect. Blurry images, handwriting, unusual fonts, low contrast, skewed pages, and complex layouts can reduce accuracy. Always verify important extracted text.
Frequently asked questions
Does this work on handwriting?
It's built for printed text — handwriting recognition is much less reliable.
What languages does it support?
The default language pack is English; accuracy drops noticeably outside of Latin-alphabet printed text since the model is trained specifically on that.
Why did some words come out garbled?
OCR accuracy depends heavily on photo quality — blur, low resolution, glare, or an angled shot are the most common causes. Retaking the photo straight-on in good light usually fixes it.
Common mistakes to avoid
Feeding in a low-resolution or blurry scan
Text recognition accuracy depends heavily on how clearly individual characters are defined in the source image. A blurry photo of a document, or a scan taken at a low DPI, will produce noticeably more recognition errors than a sharp, well-lit capture.
Not proofreading the extracted text
OCR is very good but not perfect — characters that look similar (like '0' and 'O', or 'l' and '1') are the most common source of small errors. Always skim the output against the original before using it for anything important.
Trying to OCR handwriting expecting print-level accuracy
Printed text recognition is far more reliable than handwriting recognition, since handwriting varies enormously between people. Expect more manual correction when extracting text from handwritten notes versus a printed document.