How to make a PDF smaller without wrecking it
· 4 minute read
Where the size in a PDF actually comes from, which compression settings matter, and how to tell before you send it whether you have gone too far.
A PDF that will not attach to an email is usually big for one of three reasons, and knowing which one you have tells you what to do about it. Compressing blindly either does nothing or destroys the document.
Where the size comes from
- Images. Almost always the answer. A phone photograph is several megabytes on its own, and a PDF made from photographs of pages is enormous regardless of how little text is on them.
- Embedded fonts. Usually small — tens of kilobytes — unless the document embeds several full font families rather than subsets. Then it can be megabytes before a single word is drawn.
- Everything else. Text itself is tiny. A hundred pages of pure text is smaller than one photograph.
The practical consequence: if your document is a scan, it is an image problem and compression will help a great deal. If it is a text document that is somehow large, compression will barely move it and you should look at what is embedded instead.
What compression actually does
Compressing a PDF mostly means re-encoding its images: reducing their resolution and increasing their JPEG compression. Both are lossy and neither can be undone, which is why you should always keep the original.
Resolution is the setting that matters most, and it should be chosen for how the document will be used:
- 150 DPI — fine for anything that will be read on a screen. Usually a very large saving.
- 300 DPI — what you want if it might be printed. Modest saving.
- Below 150 DPI — visible degradation. Text in scanned images starts to blur and become unreadable, and OCR will do noticeably worse on it afterwards.
Things to try before compressing
- Remove pages you do not need. Splitting out the five pages somebody asked for beats compressing eighty.
- If it came from Word, export it again with "minimum size" or "for web" selected. Producing it smaller beats shrinking it afterwards.
- If it is a scan you made yourself, rescan in greyscale rather than colour. A greyscale page is roughly a third the size and, for a printed document, loses nothing.
Check before you send
Compression damage is easy to miss because it is invisible at the zoom level people review at. Two checks take a few seconds:
- Zoom to 200% on a page with small print. Compression artefacts show up around letter edges long before they are visible at normal size.
- Try to select the text. If the document had a text layer and now does not, the compression has flattened it into an image — the file is smaller and no longer searchable.