Two modes: lossless clean-up, or rerender — both entirely in your browser
Quick answer: to compress a PDF, load it above and pick a mode. Lossless clean-up rewrites the file's structure — every pixel stays identical, but gains are modest (2.5% on a 10-page text test document we measured). Rerender compression redraws each page at your chosen DPI and JPEG quality, which is where scan-heavy files shrink dramatically at the cost of selectable text. Try lossless first, rerender if you need it smaller, and compare the reported before/after sizes.
| PDF content | Typical weight | Lossless mode | Rerender mode |
|---|---|---|---|
| Plain text pages | Tiny — a page is a few KB of instructions | Small gain (~2.5% on a 10-page test doc) | Usually grows the file — don't bother |
| Embedded fonts (full families) | Hundreds of KB to several MB | Keeps fonts intact | Drops fonts — pages become images |
| Scanned pages (photo per page) | MBs per page at scan resolution | Little gain — images untouched | Big cuts — this is the target |
| Photos and graphics | Depends on resolution | Untouched | Recompressed at your DPI + quality |
| Unused objects, bloated revisions | Varies | Some cleanup | Gone (rebuilt from scratch) |
A 300 DPI A4 scan is roughly 2480 × 3508 pixels per page — that's why a 20-page scan can outweigh a 200-page text PDF. The measured lossless figure comes from a 10-page, 400-line text document built and tested with the same pdf-lib engine this page runs: 21,865 bytes uncompressed, 21,321 with object streams.
Two engines, two philosophies. Lossless mode is pdf-lib rewriting the file: it parses every object and writes the document back using PDF 1.5 object streams — a packing scheme that stores small objects compressed inside the file. Content is preserved byte-for-byte; only the container gets tidier. Rerender mode is pdf.js plus pdf-lib: each page is drawn to a canvas at your chosen DPI, encoded as one JPEG at your chosen quality, and rebuilt into a fresh PDF, one image per page.
A scanned page is stored as a large bitmap at whatever resolution the scanner ran — often 300 DPI PNG or high-quality JPEG. Rerendering at 150 DPI replaces it with a 1240 × 1754-pixel JPEG at quality 75: fewer pixels, stronger compression, and for on-screen reading the visible difference is small. Text pages, by contrast, were already a few KB of vector instructions — rasterizing them replaces compact instructions with an image, which can make the file bigger. That's why the mode matters.
Load the PDF and the tool reports its size and page count. Start with lossless; if the number reported isn't enough, switch to rerender, set 150 DPI and quality 75, and compare. Each run shows original size, new size, and the percentage saved, so you can iterate until the file fits your target — an email attachment limit, an upload cap, a storage quota.
Take a 24-page scanned contract — every page a 300 DPI image, about 2480 × 3508 pixels. Lossless clean-up barely moves the needle: the scans are already compressed images, and repacking structure doesn't touch them. Rerender at 150 DPI, quality 75: each page becomes a 1240 × 1754 JPEG — about a quarter of the pixels — and the rebuilt PDF comes out far smaller. The table shows the exact numbers for your file, every run, before you download. Keep the original: the rerendered copy's text is no longer selectable or searchable.
Load the PDF above and try the two modes. Lossless clean-up rewrites the file structure without touching content — safe for anything, modest gains. Rerender compression redraws each page at a lower resolution as a JPEG, which is where scan-heavy files shrink dramatically. Pick the first mode that meets your size target, and compare the before/after sizes the tool reports before downloading.
Yes, but be honest about the ceiling. Lossless mode repacks the file with object streams — every pixel, glyph, and vector stays byte-identical — yet the gains are usually modest: on a 10-page text PDF built for testing, it trimmed 2.5 percent (21,865 to 21,321 bytes). Big reductions come from recompressing images, which is by definition lossy. So: lossless for documents that must stay perfect, rerender when size matters more than zoomability.
Because every page is a full-resolution photograph of paper. A 300 DPI scan of an A4 page is roughly 2480 by 3508 pixels per page, and a multi-page scan stacks those images with no re-encoding. That's why a 20-page scan can dwarf a 200-page text PDF. Rerendering such files at 150 DPI with JPEG quality around 70 usually cuts them hard, and for on-screen reading the difference is barely visible.
150 DPI at JPEG quality 70 to 80 is the workhorse setting for files that will be read on a screen — an A4 page becomes 1240 by 1754 pixels, which is sharper than most displays show. Keep 200 to 300 DPI if the file will be printed or recipients need to zoom into small print. Below 100 DPI, body text starts to look soft.
Lossless mode removes nothing you can see — it rewrites structure and can drop unused objects, but content is untouched. Rerender mode changes the document fundamentally: pages become flat images, so selectable text, embedded structure, and form fields are gone. It's a one-way door for the content, so keep the original when in doubt.
No. Both modes run in your browser — pdf-lib rewrites structure locally, pdf.js renders pages locally. The download link is built from an in-memory blob. Contracts, medical records, and financial statements never touch a server, unlike most free online compressors.