This shrinks a PDF by re-rendering each page with pdf.js at a chosen scale, re-encoding it as a JPEG at a chosen quality, and reassembling everything into a new file with pdf-lib — all inside your browser. It's most effective on PDFs that are mostly scanned pages or large images, since compression works by rasterizing content rather than touching the original vector data.
This usually happens with PDFs that are mostly vector graphics or text, or ones already well optimized. Rasterizing that kind of content into JPEG images can add size rather than remove it — this tool works best on scanned documents or image-heavy PDFs.
No. Every page is converted into an image during compression, so any selectable or searchable text in the original will no longer be selectable in the compressed output.
"High compression" uses a lower render scale and JPEG quality for the smallest file at the cost of some visual sharpness; "Low compression" keeps a higher scale and quality for the best appearance with more modest savings. "Balanced" is a reasonable middle ground for most documents.