The right level depends on what the scan is for. Start with Balanced and inspect the smallest type and any signatures at 200% zoom; move to Email & web only when the copy will be read on screen. The compressor never re-encodes an image unless it saves at least 8%, and 1-bit black-and-white scans and masks are copied as they are. An existing OCR text layer is preserved, and if the scan has none you can run OCR afterwards. Keep the untouched master for records or evidence and label the compressed copy clearly. The pages never leave your browser during any of this.
How it works
- Drop the scan and open Compress — the three sizes are computed locally
- Start with Balanced; move to Email & web for screen-only copies
- Check small print and signatures at 200% zoom, then download
Questions people ask
Why is my scanned PDF so much larger than a normal one?
Because each page is a full-resolution image rather than text. A typed PDF stores letters as compact font data; a scan stores every pixel of the paper, including the white space. A 300 DPI color scan of an A4 page is roughly eight million pixels, page after page.
Will compressing a scan hurt OCR later?
Balanced at 150 DPI is still comfortable for recognition; Email & web at 96 DPI is not ideal for small type, so run OCR before going that far, or keep the master for recognition and compress the delivery copy. An existing text layer survives compression unchanged.
Should I clean the scan before compressing it?
If pages are crooked or washed out, run Clean scanned PDF first — it deskews small rotations and lifts contrast — but note it rebuilds the pages as images and removes an existing text layer, so the order is cleanup, then OCR, then compress. Keep the original scan for archival use.