Compress PDF

Every page is rendered onto a canvas in your browser at the resolution you pick (1x or 1.5x), then re-encoded as a compressed JPEG at the quality you choose. Those images are assembled into a brand new PDF with pdf-lib. It works well because a compressed image is usually far smaller than the original page's mix of fonts, vector graphics and embedded images — especially on scans and photo-heavy pages.

Selectable text disappears because each compressed page becomes a single flattened image. Any text that was selectable in the original — real text characters positioned on the page — is rendered into pixels along with everything else, so the output PDF has no text layer to select, search or copy. This is an unavoidable side effect of the rasterize-and-recompress approach, not a bug.

Skip it for PDFs that are already small, or that are mostly text you need to keep selectable and searchable — reports, contracts, ebooks. Rasterizing those can produce a similar-sized or even larger file while destroying the text layer. This tool earns its keep on large, scanned or image-heavy PDFs, where turning vector and bitmap content into one compressed image per page saves real space.

Quality sets the JPEG compression level used for every page image — lower values shrink the file more but soften detail; around 65% is a reasonable starting point. Resolution (render scale) sets how many pixels each page is rendered at before compression — 1x keeps files smaller, 1.5x keeps text and fine detail sharper at the cost of some file size. Both settings apply to every page, so you can rerun the compression on the original file with different values.

No — the PDF is compressed on your own device. Every page is rendered and recompressed entirely in your browser using pdf.js and pdf-lib, and the result never leaves your device until you choose to download it. Nothing is uploaded, logged or stored.