How it works
Large photos inside the PDF are downscaled and re-saved as JPEG, and the file structure is rewritten compactly with object streams (pdf-lib, MIT licence). Text and vector graphics are not rasterised, so they stay sharp and searchable, and links and form fields keep working. A picture is only replaced when its new version is smaller, and if the whole file wouldn't get smaller, you're told instead of getting a copy that is no better.
PDFs that are mostly text are already small and may shrink only a little.
The three levels
- Low (best quality): pictures at most 2,400 pixels on their longest side, JPEG quality 82%.
- Medium: at most 1,700 pixels, quality 62%. Scans stay easy to read on screen and on paper.
- High (smallest): at most 1,150 pixels, quality 45%. Fine for reading on a phone or laptop.
Tick Black-and-white pictures to turn photos and scans grey as well; text colour doesn't change. That saves more space on colour scans of documents that are black and white anyway.
Example
In our test, a PDF of a 300 DPI scan of an A4 mark sheet (3.29 MB) came out at 959 KB on Low, 292 KB on Medium and 116 KB on High: 71%, 91% and 97% smaller. All 50 lines of small print on it stayed readable even on High. A PDF exported from a word processor with no pictures, by contrast, can't be made much smaller, and the tool says so.
Tips
- Remove pages you don't need first with Delete PDF pages.
- For a strict limit such as 100 KB or 200 KB, use Compress PDF to size.
- A file that won't open at all needs Repair PDF before it can be compressed.