Supported source formats
JPG, PNG, GIF, BMP, TIFF, and WebP. Drop any of those, get a UTF-8 .txt back with whatever readable text the OCR engine could find.
Why OCR an image?
- Type out a photo. Snap a picture of a printed page; get back the text without re-typing it.
- Pull from a screenshot. Code, recipes, addresses, error messages — anything you've screenshotted.
- Receipts and signs. Quick capture for expenses, addresses, transit info.
- Accessibility. Image-only content becomes readable by screen readers.
How it works
- Open the home page, drop your image in the box.
- Pick OCR (Extract Text) from the dropdown.
- Hit Convert. The image is sent to Google Cloud Vision and the recognized text is returned.
- Click Download. Output is plain text, ready to paste anywhere.
What works well
- Clean, well-lit photos of printed text.
- Screenshots of code, articles, or UI text.
- Document scans at 300+ DPI.
- Latin-script European languages by default.
What doesn't
- Heavily stylized text (logos, decorative fonts, low contrast).
- Handwritten notes — partial recognition only.
- Skewed photos taken at an angle — straighten first if you can.
- Very low-resolution images — there's no recovering detail that isn't there.
Getting better results
- Crop tightly. Less background = better signal.
- Straighten the page. A 5° tilt is fine; a 30° tilt isn't.
- Use even lighting. Shadows confuse the recognizer.
- Higher resolution helps up to a point — going from 100 DPI to 300 DPI is a huge accuracy jump; 300 to 600 is marginal.
FAQ
Does it work on screenshots? Yes — those tend to give the best results because the text is rendered cleanly.
Does it preserve layout? No. Output is plain text in approximate reading order. For tables, expect each cell on its own line.
What about handwriting? Recognition is limited; expect mistakes. Use a print-text source if you can.
Privacy? Image and resulting text are auto-deleted after one hour. See Security.