How to use Image to Text (OCR)
Extract English, Chinese and Japanese text from screenshots and scans. Up to 20 images, recognized in your browser; copy or download as TXT.
- Input
- Up to 20 JPG, PNG or WebP images, each up to 20 MB / 24 megapixels / 8,000 pixels per side, 100 MB in total
- Output
- Editable text per image, or one TXT file combined in image order
- Processing
- Processed in your browser; your input is not uploaded
- Choose one or more clear screenshots, reorder them if needed, and select an image to rotate it until the text is upright.
- Choose English, Simplified Chinese, Japanese or a mix with English, and recognize the images one by one in your browser; you can cancel, resume unfinished images or retry a single one.
- Proofread each result and copy or download it on its own, or export everything as one TXT in list order.
FAQ
Why does the first recognition take a while?
Your browser needs to download the local OCR engine and language models. Images in a batch are processed in order with the same engine, and one failed image doesn’t affect the rest. Cancelling stops the current task but keeps finished results, which you can resume or retry one at a time; closing or refreshing the page clears all files and results.
Can it read handwriting, tables or PDFs?
It works best with printed text and clear screenshots, and outputs plain text without the table layout. Handwriting, blurry images, vertical text or complex columns may come out inaccurate. For PDFs, convert them with PDF to Image first.
Are large images scaled down?
For recognition, the longest side is limited to 3,000 pixels and the total to 8 megapixels; the original file is not changed. Crop small text first, and proofread important names and numbers yourself.