What Actually Makes a PDF Smaller (It Is Not the Text)
A 40 MB PDF is almost always a pile of over-resolution images wearing a document's clothes. The two levers that shrink it, the targets that do not hurt, and what compression must never touch.
A 40 MB PDF that a portal will not accept is a specific kind of frustration: the document looks fine, prints fine, and yet the file behaves like a video. The reason is almost always the same — the PDF is carrying photographs at far more resolution than anyone will ever view them at — and the fix is more surgical than 'compress it harder'.
What is actually inside a big PDF
Open the anatomy of a heavy PDF and you usually find three things: images embedded at print resolution (300 DPI or more, sometimes straight from a scanner at full size), images stored uncompressed or in a lossless format, and occasionally the same image embedded several times. Text and vector content, by contrast, is nearly free — a hundred pages of pure text is often under a megabyte.
That asymmetry is the whole story of PDF compression. You are not compressing 'the PDF'; you are recompressing the images inside it and, optionally, lowering their resolution to something the document will actually be viewed at.
The two levers, in the right order
- 1Downsample the images. A page viewed on a screen needs roughly 96–150 DPI. A scanned A4 page at 300 DPI is 2480×3508 pixels — about four times the pixels a screen needs. Halving the linear resolution removes three quarters of the image data before quality settings even enter the picture.
- 2Re-encode the images. Scanners love lossless or lightly-compressed formats. Re-encoding those images as JPEG at quality 75–85 typically cuts them by 80–90% with no visible change at reading size.
Choosing a target that does not hurt
For documents that will be read on screens — the overwhelming majority — 150 DPI with JPEG quality around 80 is the sweet spot: text edges stay crisp and photographs look natural. Reserve 300 DPI for anything that will genuinely be printed, and accept that those files stay large.
If a portal imposes a hard ceiling ('maximum 5 MB'), treat it like any exact-size problem: search downward. Reduce quality first, then resolution, stopping as soon as the file fits, so you keep the best-looking version that satisfies the limit.
What compression should never touch
- Selectability of text. A proper PDF compressor re-encodes images and leaves the text layer alone. If a tool returns a PDF whose text you can no longer select, it rasterised the pages — avoid it.
- Vector sharpness. Logos, charts and line art should remain vectors; compressing them into pixels is a permanent quality loss.
- The page geometry. Margins, links and page boxes should survive untouched.
Local versus uploaded compression
Because a PDF's images are recompressed with standard algorithms, browser-based compression matches what a server would do — with two advantages: nothing is uploaded (relevant for contracts and records), and there is no per-file size ceiling imposed by someone else's infrastructure. The trade-off is that very large documents need a machine with enough memory to hold them while working.
Shrink the images inside your PDF while keeping text selectable and vectors sharp — in your browser, with no upload.
Open the Compress PDF tool →A worked expectation
Concrete numbers beat adjectives here, so take a typical scanned A4 page at 300 DPI — roughly 2480×3508 pixels — stored the way scanners like to store it:
| Stage | Typical page weight | What changed |
|---|---|---|
| As scanned (300 DPI, lossless-ish) | 1.5 – 3 MB | the starting point |
| Downsampled to 150 DPI | ≈ 25 – 35% of original | three quarters of the pixels removed |
| Then re-encoded JPEG q80 | ≈ 8 – 15% of original | the encoder doing the rest |
| Text-only page, any setting | ≈ unchanged | nothing expensive to remove |
Multiply by page count and you can predict the outcome before running anything: a 60-page scan at 2 MB per page is a 120 MB file that becomes roughly 10–18 MB at screen resolution, while a 60-page born-digital report barely moves at all. Knowing which of those two documents you have saves you from expecting a miracle on the wrong one.
Questions people ask
Will compression blur my text?
Not if the tool re-encodes only the images. Text stored as fonts is untouched by image compression. Blurry text after compression is the signature of a tool that rasterised whole pages; that is a different (and worse) operation.
Why is my scanned PDF so enormous?
Scanners capture at 300–600 DPI and often store each page as a large near-lossless image. That is ideal for archiving and wasteful for sharing — downsampling to 150 DPI typically removes 70–80% of the bytes on its own.
Can I compress a PDF to an exact KB?
Yes, with a search over quality and resolution the same way you would for a photo. We offer a dedicated exact-KB mode for portals with hard ceilings.
Does compression remove passwords or metadata?
Compression does not decrypt anything — a password-protected PDF must be unlocked first. Metadata is generally carried through unless a tool explicitly strips it.
Written by the Tools Stars team
We build and maintain every tool on this site, and we test them on real files before we write about them — the numbers in our guides come from those tests, not from other people's articles. If a number here changes, we re-measure and update the page. More about how we work.