Why scanned PDFs are so large
A scan is a stack of photographs, not a document. What scanner settings actually control the size, and what you can do after the fact.
In short
A scanned PDF contains one full-resolution image per page, often in colour at 300 DPI or more. Ten pages can easily be 40MB. The fix at scan time is a lower DPI and greyscale; the fix afterwards is re-rendering the pages smaller.
When you scan a document you do not get a document. You get photographs of paper, wrapped in a PDF. Nothing in that file knows it contains words.
The arithmetic
A sheet of A4 scanned at 300 DPI in colour is roughly 2480 × 3508 pixels — about 8.7 megapixels, comparable to a decent phone photo. Ten pages is ten of those. Even compressed, a few megabytes per page is normal, and 300 DPI colour is a common scanner default.
| Setting | Relative size | Good for |
|---|---|---|
| 600 DPI colour | ~4× | Photographs, artwork, archival copies |
| 300 DPI colour | Baseline | The usual default; more than most documents need |
| 300 DPI greyscale | ~⅓ of colour | Text documents with no meaningful colour |
| 200 DPI greyscale | ~⅐ of colour | Plain text you only need to read |
| Black and white / bitonal | Smallest by far | Clean printed text; poor on photos or faint pencil |
The single biggest lever is colour. Scanning a black-on-white document in full colour stores three channels where one would do, and it also stores all the sensor noise that makes flat paper compress badly.
Fix it at the scanner if you can
- Choose greyscale for anything that is only black text on white paper.
- 200–300 DPI is enough to read and to run OCR on. 600 DPI is for reproducing artwork.
- If the scanner offers a "document" or "text" mode, it usually means bitonal, which is dramatically smaller and perfectly legible on clean print.
Fixing it afterwards
Once it is scanned, the images are what they are. Re-rendering each page at a lower scale and JPEG quality is the available option, and on a scan it costs you very little — there was never any selectable text to lose. This is the one case where Dexta's compressor is unambiguously the right tool.
Before compressing, check whether you need every page. Removing the blank versos a duplex scanner produces, or the cover sheet, takes size off the file with no quality cost at all.
Mistakes worth avoiding
- Photographing pages with a phone instead of scanning. Phone cameras produce huge, unevenly-lit images that compress worse than a real scan.
- Scanning at 600 DPI "to be safe" for a document you will only read on screen.
- Compressing repeatedly. Each pass re-renders an already-degraded picture.