What makes a PDF large, and what this tool changes
In most large PDFs, the weight is in pictures: photos from a phone camera are often 4000 pixels wide, and a scanner set to 300 or 600 DPI produces a full-page image for every sheet. Text and fonts usually take only a few kilobytes per page. The tool looks at every photo, works out how large it is drawn on the page, and resizes it to the chosen resolution, so a 4000-pixel photo shown 18 cm wide becomes about 1000 pixels at 150 DPI. It is then saved as a JPEG at the chosen quality. After that, objects that nothing in the file points to are dropped (earlier versions of edited pages, unused images, page thumbnails, XMP metadata and private data left by editing programs), and the file is saved with compressed object streams. Each photo is replaced only if the new version is at least 10% smaller, so already compact images are left alone.
Choosing Low, Recommended or Strong
Low keeps photos at 200 DPI with JPEG quality 85%: pictures stay crisp on paper and the saving comes mostly from very large camera photos. Recommended uses 150 DPI and quality 72%, which looks sharp on phones, laptops and office printers and suits forms, reports and ID copies sent by email. Strong uses 100 DPI and quality 55% for the smallest file, for example when a portal accepts only a few megabytes; small print inside photos, such as the numbers on a scanned receipt, may become soft, so open the result and check it before you send it. You can run the same file again with another level, because your original is never changed.
Why some PDFs barely shrink
A PDF that is mostly typed text, such as a contract exported from Word, is already compact, and there is little to gain; the tool says so plainly instead of showing an invented number. Black-and-white scans are often stored in special fax-style formats (CCITT or JBIG2) that are smaller than any JPEG, so they are kept as they are. Print-ready files with CMYK photos, colour-indexed images and vector drawings are also left unchanged, because converting them could shift colours or blur lines. When the finished file is not smaller than the one you gave, you get your original back with a note, never a bigger file.
Checks on every file, and what to do first
Before a compressed PDF is offered for download, it is opened again and its page count is compared with the original; if anything does not match, you get the original instead. Pages, text, links, bookmarks and form fields are not touched, so text stays selectable and searchable. In our tests, compressed files opened normally in Poppler (pdftotext) and LibreOffice with the same pages and the same text. A password-protected PDF has to be unlocked first with Unlock PDF. Compressing rewrites the file, so a digital signature will show as no longer valid: compress before signing, not after.
Privacy, limits and speed
Your PDFs never leave this browser tab. The file is read with the open-source pdf-lib library and the photos are re-encoded with the image encoder built into your browser, so nothing is sent to NeedMyTools or anyone else, and the tool keeps working if your connection drops after the page has loaded. You can add up to 20 files of up to 150 MB each per run. On our test computer (4 processor cores, Chromium), a 6-page PDF of 4000 x 3000 phone photos (9.5 MB) took about 1.3 seconds and a 24-page one (38 MB) about 4.6 seconds; your time depends on your device. To bring a very large file under a strict limit, split it into parts first, or compress single photos with our image compressor before you make the PDF.