PDF Compression Guide

A PDF is a container of objects: text streams, embedded font subsets, vector paths, and images. Text and vectors are already Flate-compressed when the file is written, which is why a 60-page contract can be under 1MB. Images are the variable — a single 12-megapixel photo can outweigh the rest of the document.

Expected results by document type

Scanned documents: 70-90% smaller. A 48MB scan typically lands near 4MB at 150 DPI with no practical readability loss.

Text-heavy reports: 5-20% smaller. Large drops here mean pages are being rasterized and quality will suffer.

Slide decks: 40-70% smaller, because they carry full-resolution photography behind small frames.

Print-ready artwork: stay at 300 DPI; lower settings show banding once on paper.

Choosing a DPI

96-150 DPI for email and screen reading, 200 DPI for casual printing, 300 DPI when print quality matters. Above 300 DPI adds size with no visible benefit for office printing.

A workflow that avoids rework

Decide the destination first. An upload or email limit is a hard target, while print quality is a visual requirement.
Compress once at a moderate setting and inspect the result. Repeated JPEG passes stack artifacts and save less each time.
Check signatures, chart labels, fine lines, and footnotes because they show damage before ordinary body text.
If the result is still too large, split the document rather than making every page unreadable.
Add signatures, passwords, or watermarks after the size is settled because compression rewrites the file.

Why compression sometimes makes a file larger

Re-encoding an image that was already aggressively compressed can add bytes instead of removing them. Text and vector graphics are usually compact already, and rebuilding them offers little benefit. When the output is larger, keep the original rather than running another pass. A second pass cannot recover detail removed by the first one.

Browser and document limits

The saved file size is not the same as the memory needed to process it. A compressed scan expands into full page bitmaps during rendering, so a 100MB document can require several hundred megabytes of working memory. Desktop browsers handle these jobs more reliably than phones. Password-protected files must be unlocked before their content streams can be inspected, and digitally signed PDFs should be compressed before signing because rewriting invalidates the signature.

Common questions

Why did my PDF barely get smaller? Because it was already efficient. Text-only documents have little left to remove without converting pages into images.

Is there a size limit? Everything runs in your browser tab, which is memory-bound. Files up to about 300MB are reliable on a desktop; mobile browsers cap out lower.

Will compressing break a signature? Yes. A digital signature covers the exact byte layout, so sign after compressing, not before.

Does compression remove metadata? Rebuilding may drop incidental producer data, but compression is not redaction. Any information visible on a page remains visible.

Are files uploaded? No. PDF and canvas operations run in the current browser session, and the result is created as a local download.