Compress PDF
Make a PDF smaller by re-encoding its images. Text stays text and stays selectable. The file is processed in your browser and never uploaded.
Drop a PDF here
Scans and PDFs full of photos shrink the most.
How it works
Open a PDF and it is compressed straight away at the level you picked. The page shows the size before and after, how many images were re-encoded, and a Download button. Pick another level and the same file is compressed again from the original, so you can compare without reopening it.
What actually shrinks. In most large PDFs the weight is in the pictures: scanned pages, photos, screenshots pasted into a report. Each eligible image is decoded, scaled down so its longest side is at most 2400 px (Light), 1600 px (Balanced) or 1100 px (Strong), and written back as a JPEG at quality 80%, 65% or 50%. An image is only replaced when the new version is smaller; a small image stays at its size and is just re-encoded. An A4 page scanned at 300 dpi is about 3500 px tall, so Balanced keeps roughly 135 dpi, which still reads well on screen and prints fine for text.
What does not change. Pages are not rasterized. Text stays text, so it can still be selected, searched and copied, and vector drawings, fonts, links, bookmarks and form fields are copied as they are. That is also why a PDF made only of text and vectors (an invoice exported from a program, say) shrinks little: there is nothing heavy to re-encode.
Which images are skipped. CMYK images from print workflows, palette (indexed) images, images with a colour-key or stencil mask, and black-and-white scans stored as CCITT or JBIG2 (already tiny) are left alone, as are images under 10 KB. Grayscale images come out as RGB JPEGs that look the same. Images with a transparency mask are re-encoded and keep their mask.
Lossless clean-up. Objects that no page refers to any more, the leftovers of a PDF edited and saved many times, are removed, and the file is written with compressed object streams. If the result is still not smaller than the original, the tool says so and does not offer a larger file.
Privacy. The PDF is read with the File API and processed by pdf-lib and the browser's own image decoder inside this tab. Nothing is uploaded, which matters for contracts, payslips and scanned IDs. Password-protected PDFs cannot be opened: remove the password first. To merge, reorder or delete pages before compressing, use PDF Toolkit.
Frequently Asked Questions
How much smaller will my PDF get?
- It depends on how much of the file is images. A 300 dpi scan of an A4 page is about 3500 px tall; Balanced scales it to 1600 px, roughly a fifth of the pixels, so scans and photo-heavy PDFs shrink the most. A PDF that is mostly text and vector graphics may shrink by only a few percent, because text is already compact and is left untouched.
Does compressing a PDF make the text blurry?
- No. Text and vector drawings are copied unchanged, so they stay sharp and selectable at any zoom. Only the embedded images are downscaled and re-encoded. In a scanned PDF the text is part of the image, so there Strong can make small print softer; use Light or Balanced for fine print.
Which compression level should I use?
- Balanced for most files: images up to 1600 px at JPEG quality 65% look fine on screen and on paper. Light keeps images up to 2400 px for photos you may zoom into. Strong caps them at 1100 px, for when the file has to fit an email or upload limit.
How do I compress a PDF to send it by email?
- Open it here with Strong selected and check the size shown after compression. Most mail services cap attachments at 20–25 MB. If a scan is still too large, remove the pages you do not need with PDF Toolkit and compress again.
Why did my PDF not get smaller?
- Because it has nothing heavy the tool can re-encode. Text-only PDFs, black-and-white scans stored as CCITT or JBIG2, and images in CMYK or palette colour are kept as they are. When the result is not smaller, the tool keeps the original and does not offer the download.
Can I compress a password-protected PDF?
- No. An encrypted PDF has to be decrypted with its password before its images can be read, and the tool does not ask for passwords. Remove the protection in your PDF reader first, then compress the unprotected copy.
Is my PDF uploaded to a server?
- No. The file is read and compressed by JavaScript in your browser tab, using pdf-lib and the browser's built-in image decoder. DevTools → Network shows no request carrying the file.
Compression levels and what they do to images
| Level | Longest image side | JPEG quality | Good for |
|---|---|---|---|
| Light | 2400 px | 80% | Photos and detailed scans you may zoom into |
| Balanced | 1600 px | 65% | Most documents, on screen and printed |
| Strong | 1100 px | 50% | Email attachments and upload size limits |
Text, fonts and vector graphics are never re-encoded at any level. An image is only replaced when the re-encoded version is smaller than the original.