Tested guides · Original examples
Before compressing a PDF: compare scanned and text pages
Two one-page A4 files can have very different internal structures. Compare compression on two original examples before deciding to lower quality further.
Maintained by:Finl Tools · Updated: · Example dates and methods appear below
Open PDF compressionFind out what the file contains
Open the file in PDF inspection and check each page’s text layer and image objects. A scan usually stores the whole sheet as an image; a text page stores glyphs and layout instructions. These structures respond differently to compression settings.
The scanned example contains one image and no extractable text. The text example contains 940 non-whitespace characters and no image objects. This is a structural distinction, not a measure of content accuracy.
Compare two runs from the same input
Download the original scan. Select it afresh for each run, use quality 70 and 40 respectively, and choose PDF output. Do not feed the first compressed output into the second run; that would introduce repeated encoding into the comparison.
Check small text, fine table lines and gray texture in both outputs. Similar-looking thumbnails do not guarantee clear printing. View at 100% in a reader and try printing the key pages you need to deliver.
Interpret the measured differences
The original scan is 709772 bytes. Quality 70 produces 356760 bytes and quality 40 produces 256435 bytes. All three are one A4 page, with no added searchable text layer.
The text example drops from 5390 bytes to 1395 bytes, mainly through document-structure and content-stream optimization. It has no images, so the reduction cannot be attributed to the image-quality setting or generalized into a compression ratio for all text PDFs.
Know when to stop compressing
Stop once the file meets the recipient’s size limit and important information is clear. Re-encoding an already compressed image may change little or increase size; encoding overhead varies by format and affects small files.
If still over the limit, consider removing irrelevant pages or splitting as permitted. Compression is not OCR and does not automatically fix skew, blur or missing text. Keep the original and avoid using low-resolution output as the source for later editing.
Download the input and actual outputs
All files are synthetic examples created by this site without real business information. Download them to repeat the operations.
Example tested on 2026-09-26. PDFs used the site API (pypdf 6.19.0 / PyMuPDF 1.28.2); images used Canvas encoding in Chromium 153. Other browsers or versions may produce different file sizes.
| File | Actual bytes | Download |
|---|---|---|
| Original scan | 709772 B | Download file |
| Scan at quality 70 | 356760 B | Download file |
| Scan at quality 40 | 256435 B | Download file |
| Original text PDF | 5390 B | Download file |
| Compressed text PDF | 1395 B | Download file |