Pull the text out of a screenshot, scan or photo and edit it as plain text. Recognition runs inside your browser — the image is never uploaded.
No. The OCR engine is Tesseract compiled to WebAssembly and it runs in your browser tab, which means confidential documents stay on your device.
The engine and the language data have to download the first time — a few megabytes depending on the language. They are cached afterwards, so later scans start right away.
Clean screenshots and flat scans of printed text are usually 90%+ accurate. Angled photos, low contrast, unusual fonts and handwriting are much weaker. Every result shows a confidence score so you know when to proofread.
Crop tightly to the text, keep the page straight rather than skewed, avoid shadows, and select the correct language — a mismatched language is the single most common cause of garbled output.
Not directly. Convert the PDF page to an image first, then run it through here.