How to Create a PDF From Scanned Documents

Scanner settings decide whether you get a clean 2 MB document or an unreadable 90 MB one. Here is what each setting actually does.

Scanning is one of those tasks where the defaults are usually wrong in both directions: too high for ordinary paperwork, too low for anything with fine detail. A little deliberate setting-up produces better documents and much smaller files.

Resolution

  • 200 dpi — adequate for large, clean printed text. Noticeably rough on small print.
  • 300 dpi — the right default for almost all documents, and the usual minimum for OCR.
  • 600 dpi — for fine detail, small footnotes or anything being archived permanently. Files grow roughly fourfold over 300 dpi.

Colour mode

This choice affects size more than resolution does. Black and white (bitonal) is dramatically smaller and perfectly good for plain printed text, but it destroys signatures, stamps and pencil annotations, which turn into blotches. Greyscale keeps them and stays reasonably compact. Full colour is only worth it when colour carries meaning — a highlighted clause, a colour-coded chart, a company seal.

Before you scan

  1. Remove staples and paperclips, and flatten folds — a crease casts a shadow that reads as a grey band.
  2. Set the correct source orientation so pages do not arrive sideways.
  3. Clean the glass or the feed rollers. A single speck produces an identical mark on every page of a hundred-page batch.
  4. Scan a single test page and inspect it at full size before committing to the batch.

Scanning to PDF directly, or via images

Most scanners can output a multi-page PDF, which is convenient and usually the right choice. Scanning to individual images gives you more control: you can discard failed pages, straighten and crop them, compress them properly, and only then build the PDF. For an important or long batch, the image route generally produces a better document.

A scanned PDF has no text layer. It cannot be searched, copied from, or read by a screen reader until OCR has been applied.

Cleaning up afterwards

Duplex scanning a single-sided stack produces a blank page after every real one — delete them in a single pass. Check for rotated pages and fix the rotation in the file. If the result is still large after all that, compress, and check the smallest text before accepting the saving.

Conclusion

300 dpi, greyscale unless colour matters, a test page before the batch, and a cleanup pass afterwards. That combination handles the overwhelming majority of scanning jobs without producing files anyone has to complain about.