How to compress a PDF without losing quality
Which parts of a PDF can shrink losslessly, which cannot, and how to pick a compression level that stays visually sharp. An honest guide — including when quality loss is unavoidable.
The honest answer first
“No quality loss at all” is only true for some PDFs. Text, fonts and vector graphics can be re-compressed losslessly — they stay pixel-identical. Photos and scans, which are what makes most PDFs big, can only shrink meaningfully through lossy re-encoding.
The realistic goal is therefore visually lossless: reduce image data to the point where the difference is invisible at normal reading size, and never touch the text layer.
What each compression level actually does
A standard level downsamples images to roughly screen resolution and re-encodes them at a quality where JPEG artifacts are not noticeable in normal reading. Stronger levels go below that — fine for drafts and email, not for print.
For print, keep images at 150–300 DPI. For screen reading and email, 96–120 DPI is usually indistinguishable. Text is unaffected either way, because real text layers are never rasterized.
How to check the result
Zoom to 200% on a photo-heavy page and compare edges and gradients with the original. Paplume shows an estimated output size before you run (computed locally, no quota consumed), and if compression would gain nothing it says so and consumes no quota. Processing is local — the file is never uploaded.
Step by step
- 1
Open the compressor
Free during Beta, no sign-up; the file is processed locally in your browser.
- 2
Start with the standard level
Check the pre-run size estimate. Standard aims for visually lossless at reading size.
- 3
Compare at 200% zoom
Open the result next to the original on an image-heavy page. If you can see a difference, use a lighter level.
- 4
Only go stronger when size demands it
For hard portal limits use target-size mode — it tells you the reachable floor honestly.
FAQ
Is truly lossless PDF compression possible?
For text/vector content, yes. For photos and scans, meaningful savings require lossy re-encoding — the goal is to keep the loss invisible at reading size.
Will my fonts or text sharpness change?
No. Real text layers are never rasterized by compression. Only scanned "text" (which is an image) can soften.
Does the estimate consume my quota?
No — the estimate is computed locally by sampling pages and costs nothing. Failed compressions refund the quota automatically.