Reads the text in an image. The picture is prepared locally first — flattened, converted to grey, contrast-boosted and scaled so small scans are legible — then recognised by Tesseract running in your browser. The engine and the language data are downloaded from the CDN the first time you use a language and cached by the browser afterwards, so the first run needs the network and takes longer; the image itself is never uploaded. Output can be plain text, one line per row, or TSV with the confidence and box of every word.
Combine two with a plus, for example eng+ind, when a page mixes languages.