Why a PDF will not compress, and what makes PDFs large
· 6 min read
Pictures, fonts, drawings, attachments and leftovers: what takes the space in a PDF, a one-minute test to tell which one is yours, and what to do about each.
A PDF is a container
A PDF holds objects: pages, the text drawn on them, the fonts the text uses, the pictures, the drawings, and a table that says where each object is. Most of these are already stored compressed. So compressing a PDF is not like zipping a folder; it is finding the objects that are larger than they need to be and storing them differently. When none of the objects is like that, the file does not shrink, and no tool can change that without changing what the document contains.
The five things that take space
- Pictures
- Scans, photos, screenshots and logos. By far the usual cause: a scanned page at 300 dpi is 300 to 800 KB, a photo from a phone 2 to 5 MB. Stored at the size they are drawn and at a sensible quality, they take a fifth of that or less.
- Fonts
- A document should embed only the letters it uses (a subset). Some exporters embed the whole font, and a font for an Indian script or Chinese runs to several megabytes. A text-only PDF of 4 MB is almost always this.
- Drawings
- Maps, CAD exports and charts with thousands of tiny paths are vector drawings. They are already compact for what they are; the only way to make them smaller is to simplify them in the program that made them.
- Attachments
- A PDF can carry other files inside it: a spreadsheet behind a report, a 3D model, the original Word file. They are invisible on the page and count in full.
- Leftovers
- Every save in some programs appends to the file instead of rewriting it, so deleted pages and old versions are still inside. Pictures used on several pages may be stored once per page, and some files keep a thumbnail of every page.
A one-minute test to tell which one is yours
Open the file in Compress PDF and run it twice, once at Lossless and once at Extreme. Lossless touches no picture: it rewrites the file cleanly, packs the small objects and drops what nothing refers to. Extreme re-encodes every picture at 72 dpi. Then read the two results together:
- Lossless shrinks it a lot
- Leftovers. The clean copy is your answer, and nothing was lost.
- Extreme shrinks it a lot
- Pictures. Pick the level between the two that still looks right, or type a target size.
- Neither moves it
- Fonts, drawings or attachments. The remedies are below; none of them is a compression setting.
- You cannot select the text
- The pages are scans, so it is pictures all the way; the Extreme result shows the floor.
| File | Original | Lossless | Extreme |
|---|---|---|---|
| Text report, 3 pages | 9.6 kB | 0% smaller | 0% smaller |
| Form with fields, 1 page | 18 kB | 21% smaller | 21% smaller |
| Scanned invoice, 2 pages | 410 kB | no change | 85% smaller |
| Photo brochure, 6 pages | 2 MB | no change | 89% smaller |
The report hardly moves at either level, because it is text and a few small drawings. The form loses a fifth at every level, Lossless included: that is packing, not pictures. The scan and the brochure lose nothing at Lossless and most of their size at Extreme, because they are pictures.
What the tool leaves alone on purpose
Some pictures are skipped even at Extreme, and a file made of them will not shrink here: pictures stored as JPEG 2000, JBIG2 or fax (CCITT) encodings, which are already compact and would get larger as JPEG; pictures with a transparency mask or a stencil mask, which JPEG cannot carry; CMYK JPEGs meant for print; and tiny pictures, where the overhead of re-encoding outweighs the gain. The tool never turns text or drawings into pictures either. A blurry, unselectable page is worse than a large one.
What to do for each cause
- Pictures: Compress PDF at Strong or Extreme, or a target size; tick Turn pictures gray for black-and-white pages. The guide on getting under 200 KB goes through it.
- Fonts: export the PDF again from the original document with the smallest-size or online option, which subsets fonts, or print it to PDF. If you only have the PDF, the fonts can be dropped only by turning pages into pictures, which is rarely worth it.
- Drawings: simplify in the program that made them, or, for a page that only needs to be looked at, save it as a picture with PDF to JPG and wrap it with JPG to PDF.
- Attachments: remove them in the program that added them; a PDF saved by any tool on this site is written afresh, and carries over only the pages.
- Leftovers: the Lossless level, or simply saving a copy from any tool here, writes a clean file. Old versions and deleted pages are gone from the copy, which is also worth knowing before you share one.
- Too many pages: Split PDF into the parts people actually need. Two files of 1 MB go where one of 2 MB does not.