Extract readable text from a screenshot
A practical OCR workflow for clean source images, correct language selection, and accurate copy.
Prepare the screenshot
Image to Text uses Tesseract OCR locally. It reads characters from pixels; it does not understand whether a copied amount or address is correct. Upload a JPG, PNG, WebP, or BMP under 15 MB. This input does not accept a PDF directly: export the page you need as an image first.
Crop away browser menus, straighten tilted text, and keep letters large enough to inspect. Increasing the size of a blurry thumbnail cannot restore missing character detail. For a multi-column page, extract one column at a time to reduce reading-order errors.
Extract, compare, then copy
- Select the image and choose the OCR language that matches the document.
- Choose Extract text. The first run downloads runtime and language files; the image itself stays in local processing.
- Compare the output line by line with the screenshot. Correct mistakes in the result editor before exporting.
- Use Copy text or Download text for the corrected result.
A useful review example
Suppose a fictional screenshot reads:
Subtotal: $180.00
Due: 20 October
Check that the output has not changed 014 to O14, lost the decimal point, or moved the date into a different line. This is an illustrative checking exercise, not a claimed OCR result. For invoices or contact details, these small errors matter more than whether a paragraph looks fluent.
OCR output is plain text. It does not preserve a complete spreadsheet, editable document layout, or trustworthy table structure. Handwriting, decorative type, curved pages, and low contrast often need manual transcription.
When nothing readable appears
Confirm that the picture contains visible text, choose the correct language, and try a tighter crop. If the OCR library fails to load, refresh once and check your connection. Do not send private documents through the contact form when reporting a bug; describe the problem or use a non-sensitive sample.