Compress PDF

Drop PDFs and pick a level. Light rewrites the file structure and drops page thumbnails without touching images; Balanced caps photos at 1600 pixels on the long side and re-encodes them as JPEG at quality 80; Strong caps everything at 1100 pixels at quality 62, fine for reading on screen. Each image is replaced only if the result is actually smaller. Before and after sizes are shown per file, with a plain account of what was re-encoded, what was left alone and why, and an optional switch to strip the document’s title, author and XMP metadata too.

What this honestly can and cannot do. A PDF is mostly two things — content streams, which are already compressed, and images, which usually are not compressed well. This tool decodes, downscales and re-encodes the images; text, fonts and vector graphics are left exactly as they are, so text stays selectable and searchable. A scanned document or a photo-heavy deck typically shrinks by half or more; a text-only PDF barely moves, and the tool says so instead of pretending. CMYK and JPEG 2000 images, stencil masks and palette images are left untouched rather than risk a colour shift.

The work happens in a background worker in your browser using the same image codec the browser uses for the web. Nothing is uploaded. Every other “compress PDF online” result sends the file to a server first — and the files people compress to get under an email limit are exactly the ones they would least want on someone else’s disk.