Compress vs Optimize: what each level actually changes
Light, Balanced, Strong — three levels with different trade-offs.
Heads up: we don't yet ship a built-in compress tool ourselves (pdf-lib's API doesn't support compression — a server-side qpdf+ghostscript pipeline is on the roadmap). The conceptual breakdown of Light/Balanced/Strong below applies to whatever compression tool you do use (Adobe Acrobat Online, iLovePDF, SmallPDF). We'll restore this post's tutorial framing when the real tool ships.
PDF compression is mostly image compression. The text in your file is already vector data — it shrinks barely at all. The savings come from re-encoding embedded images at lower JPEG quality and downsampling them.
Light: ~80% JPEG quality, no downsample
Re-encodes images at 80% JPEG quality, leaves DPI alone, strips unused fonts. Result: ~20% smaller, no visible quality loss. Use for print-bound documents and archival copies.
Balanced: ~60% JPEG quality, 200 DPI
The default, calibrated for screen reading. Cuts file size 40-60%. Quality is indistinguishable from the source at fit-to-width. Best for email and on-screen reading.
Strong: ~40% JPEG, 150 DPI
Aggressive compression for upload caps and mobile-first reading. Cuts size 60-80%. Quality is visible if zoomed past 100%. Don't use for print-bound content.
Posts authored collectively by the pdfcraft ai team — release notes, infrastructure updates, and editorial content where no single author is the right attribution. Read more →