Compress PDF

Compress PDF makes a PDF file smaller inside your browser, and tells you plainly how much smaller — or that it could not. It is for the everyday problem of a PDF too big to email or upload to a form, especially a scanned document or a file full of photos, when you would rather not send a private document to someone else's server.

Be clear about one thing first: compressing in a browser is weaker than the big server-side tools, which run programs like Ghostscript that can re-encode each image inside a PDF individually. This tool does not pretend otherwise. It offers two honest modes and always shows the original size, the new size and the percentage change. If the result is not smaller, it says so and recommends keeping your original.

Lossless clean-up is designed to change nothing you can see. The PDF is opened with pdf-lib and rewritten keeping only what the document still uses: objects nothing refers to any more (leftovers from editing, unused fonts and images, old versions kept by incremental saves) are dropped, any stream stored without compression is compressed with Deflate — the same lossless method as ZIP — and small objects are packed into compressed object streams. Text stays selectable, images keep every byte, and bookmarks, links and form fields are kept. Many PDFs are already stored efficiently and shrink only a few percent or not at all; files that have been edited and re-saved many times, or exported by software that skips compression, can shrink a lot. One caution: pdf-lib reads PDFs leniently, so if the original is malformed, parts it cannot parse may be missing from the result — open the compressed file and check it before you delete the original.

Strong mode rasterises: pdf.js, Mozilla's PDF renderer, draws each page as a picture at the resolution you choose (72, 100, 110 or 150 DPI; 110 by default, which turns an A4 page into 909 × 1286 pixels), saves it as a JPEG at the quality you set (70 by default), and a new PDF is built with each picture filling a page of exactly the original size. This is what makes big scans and photo-heavy PDFs dramatically smaller. The price is real: text becomes part of the image, so it can no longer be selected, searched or copied, links and form fields stop working, bookmarks are gone, and fine print can blur at low resolutions. Use it for scanned documents, not for text documents you still need to search. For a PDF that is mostly text, strong mode often makes the file bigger, and the tool will tell you.

Password-protected or permission-locked PDFs are refused in both modes. Files up to 200 MB are accepted.

Your file never leaves your device: it is read, compressed and handed back inside the browser tab. You can check in your browser's developer tools (Network tab) that no request carries the PDF — the site loads its own scripts and analytics, never your documents.

1. The file's first bytes are checked for the PDF signature, and pdf-lib opens it; encrypted and damaged files are refused with a specific message (in both modes). 2. Lossless: every stream stored without a compression filter (other than XMP metadata, which is left readable) is Flate-compressed if that makes it smaller; the decoded bytes are identical. 3. Lossless: starting from the document catalog and info dictionary, every object still referenced is marked; everything else is deleted. The file is saved with compressed object streams and a cross-reference stream. 4. Strong: each page is rendered by pdf.js at the chosen DPI (pixels = points × DPI ÷ 72, reduced only for pages that would exceed 16 megapixels or 8,192 pixels per side), painted on white, and encoded as a JPEG at the chosen quality by the browser. 5. Strong: a new PDF is built with pdf-lib, one page per JPEG, each page the original page's displayed size in points with the JPEG drawn to fill it. The JPEG bytes are embedded as-is. 6. The original and new sizes and the percentage change are shown; a result that is not smaller is labelled as such.