How PDF Compression Actually Works

A plain-English look at what happens inside a PDF when you shrink it — and how to do it without ruining quality.

"Compress a PDF" sounds like one simple action, but under the hood there are several very different techniques at play, and they don't all preserve your document equally. Understanding the difference is the key to shrinking a file as much as possible without turning your crisp text into a blurry mess.

First, what's actually inside a PDF?

A PDF is a container. Inside it you'll typically find a mix of several kinds of objects:

• Text, stored as characters plus references to fonts — this is usually tiny.
• Vector graphics (lines, shapes, logos) described as mathematical paths — also small and infinitely scalable.
• Raster images (photos, scans, screenshots) stored as grids of pixels — these are almost always the biggest part of a large PDF.
• Embedded fonts, metadata, annotations, form fields, and sometimes attachments or duplicate resources.

When a PDF is surprisingly large, the culprit is nearly always images — high-resolution scans or photos that were dropped in at full size. That single fact drives almost every compression strategy.

Lossless vs lossy: the fundamental split

There are two philosophies of compression, and a PDF tool may use either or both.

Lossless compression

Lossless methods make a file smaller while keeping every bit of the original data recoverable — nothing is thrown away. They work by spotting redundancy: repeated patterns, long runs of identical pixels, or duplicate objects that can be stored once and referenced many times. Removing an unused font, de-duplicating an image that appears on every page, or zipping the internal data streams are all lossless wins. The catch: lossless savings on an already-efficient file are modest — often single-digit to low double-digit percentages.

Lossy compression

Lossy methods achieve dramatic size reductions by permanently discarding information your eye is unlikely to miss. The classic example is JPEG image compression, which throws away fine detail and subtle color variation. Push it gently and the image looks identical; push it hard and you get visible blocky "artifacts." Lossy compression is where the big savings live — but it's a one-way street. Once detail is gone, it's gone.

The most common technique: re-rendering pages as images

A very effective and widely-used approach — and the one CrunchyPDF's Compress tool uses — works like this:

1. Each page is rendered (drawn) onto a canvas, exactly as it would appear on screen.
2. That canvas is then re-encoded as a JPEG at a quality level you choose.
3. The compressed page-images are reassembled into a new PDF.

Because JPEG is extremely good at squeezing photographic and scanned content, this can cut a file by 40–80%. The quality slider directly controls how aggressive the JPEG step is: higher quality keeps more detail and a bigger file; lower quality shrinks harder with more visible softening.

Other techniques you'll encounter

• Downsampling images — reducing a 600 dpi scan to 150 dpi keeps it sharp on screen while cutting pixel count (and size) enormously.
• Re-compressing existing images — converting a bulky PNG screenshot to a JPEG, for instance.
• Font subsetting — embedding only the characters actually used instead of an entire typeface.
• Object de-duplication and stream compression — the lossless housekeeping mentioned above.
• Stripping metadata and unused objects — small but free savings.

How to choose the right setting

A practical workflow

Start at a middle quality setting, compress, and open the result. If it looks great, try a lower setting and compare — you might save even more. If text or fine lines look fuzzy, step back up. Because the whole process in CrunchyPDF runs in your browser, you can experiment freely: nothing is uploaded, and your original file is never touched.