When OCR helps a scanned PDF — and when it wastes your time

OCR means the computer looks at pixels and guesses letters. Pi2PDF.com runs that in the browser and currently aims at printed English. The first run may download a language model. That download is a library, not your document.

If the PDF already has selectable text, use PDF to Text instead. OCR on a digital letter just adds mistakes.

A scan that works

Lay the page flat. Fill the frame. Avoid glare. Dark text on white paper, 300 DPI or a sharp phone photo, is the happy path. Proofread names, numbers, and headings. OCR is a draft, not a notary.

A scan that fails

Cursive, stamps over text, tables shot at an angle, and low-contrast grey-on-grey forms will come out as junk. Fix the photo first, or type the few lines you need.