How to compress a PDF without losing quality (the real answer)

"Compress PDF without losing quality" is half a lie. Any compression on a PDF with images involves lossy re-encoding somewhere — the question is how much quality loss, in which parts, and whether you can see it.

Here's the truth in three levels.

What actually compresses a PDF

PDFs are already compressed internally — text streams use FlateDecode (DEFLATE), images use JPEG or JBIG2. The "compress PDF" tools you find online do some combination of these four things:

1. Re-encode embedded images at lower JPEG quality. The biggest win. A 4000×3000 photo at quality 95 is 1.8 MB; at quality 72 it's 420 KB. Most people can't tell the difference at screen resolution. 2. Downsample images above target DPI. A 300-DPI scan printed at 150 DPI is 4× fewer pixels. Huge savings on scanned documents. 3. Strip metadata, thumbnails, unused objects. PDFs accumulate cruft — author fields, Acrobat revision history, orphan image references. Small wins individually, meaningful on complex files. 4. Rebuild the cross-reference table with object streams. Pure savings, no quality impact. Just shorter bookkeeping.

The three-level trade-off

On SnapPDF's compress tool, the three levels do:

• Gentle: items 3 and 4 only. Zero visible quality change. Savings: 5–25%. Safe for legal docs, print-ready files, archival.
• Balanced: all four. JPEG quality 90, downsample to 150 DPI. Savings: 40–60%. Invisible at screen resolution, tiny loss on 4K displays and large prints.
• Aggressive: all four, harder. JPEG quality 72, downsample to 96 DPI. Savings: 60–72%. Visible on close inspection but fine for email attachments and preview-only sharing.

What to avoid

• Re-compressing an already-compressed PDF. Running compress twice re-encodes images that were already quality 72 down to quality 72 of an already-degraded image. Quality craters. If a PDF didn't shrink on the first pass, it won't shrink on the second.
• Using "compress" tools that actually just print-to-PDF. Several free tools rasterize every page to image, then embed that image in a new PDF. The output "compresses" massively but the text becomes an image — unsearchable, unselectable, accessibility-broken. Watch the output: if you can't select text in the result, you got scammed.
• Converting PDF to Word and back. Loses formatting, re-encodes fonts, doubles your problem.

When compression isn't the answer

If your PDF is 50 MB but you need it under 25 MB for email, try these first before compression:

• Split the file. /tools/split-pdf into 3 parts, send separately.
• Share via link. Drop it in Drive/Dropbox, send the link. No compression needed.
• Send the source instead. If the PDF came from an exported slide deck, send the deck.

Try it

• /tools/compress-pdf — drop a file, see the exact savings before downloading
• /tools/merge-pdf — combine first, compress second for maximum savings
• API docs — call it from code, same three levels