How to use the PDF Compressor
- Choose a PDF.
- Pick a Compression preset: Lossless, Balanced, Strong or Extreme.
- For scans of printed text, tick Greyscale to save more space.
- Press Compress PDF.
- Compare the new size, the original size, the percentage saved and the method, then download the copy ending in -compressed.pdf.
What it does and when to use it
Large PDFs are slow to email, fill up cloud storage and get rejected by upload forms. Why a PDF is large decides what can be done about it, and that is why this tool offers presets rather than a single button.
A PDF made on a computer, such as a Word export or a report from accounting software, is mostly text and vector graphics. These files are often bloated by duplicated fonts or uncompressed internal data, and a lossless rebuild can trim them without changing a single visible pixel.
A scanned PDF is different. Each page is a photo, often stored at a much higher resolution than anyone needs to read it. A lossless rebuild barely touches such files. The only way to shrink them is to redraw the pages at a lower resolution, which is what Balanced, Strong and Extreme do.
The four presets
| Preset | What happens | Text selectable? | Best for |
|---|---|---|---|
| Lossless | Internal structure repacked; nothing redrawn | Yes | Word exports, reports, forms |
| Balanced | Pages as JPEG at 150 DPI, quality 72% | No | Scans you still want to look good |
| Strong | Pages as JPEG at 110 DPI, quality 55% | No | Large phone scans for email |
| Extreme | Pages as JPEG at 72 DPI, quality 40% | No | Archival copies where size matters most |
Every preset runs the lossless pass first. The image presets are used only if they actually produce a smaller file than that.
How it works
The work happens in your browser. For the lossless pass, the tool rebuilds the PDF with pdf-lib, using compressed object streams, and also runs qpdf compiled to WebAssembly to recompress internal streams. It keeps whichever result is smallest.
For the image presets, pdf.js renders every page at the preset’s resolution, optionally in greyscale, and each page is saved as a JPEG at the preset’s quality inside a new PDF. Each page keeps its size, but its content becomes one picture. A notice in the result reminds you that text, links and form fields no longer exist in that copy.
Worked examples
A 1.2 MB report exported from Word. Lossless may bring it down noticeably if the export embedded fonts inefficiently, with nothing re-rendered and the text still selectable. If the result says there was little to save, the size is coming from images inside the report.
A 12 MB phone scan of a ten-page document. Lossless saves almost nothing. Strong redraws the pages at 110 DPI, which for A4 is about 909 × 1286 pixels per page, and the file usually shrinks to a small fraction of its original size. Tick Greyscale for plain printed text to save more.
A scan that must stay sharp for printing. Use Balanced. At 150 DPI an A4 page is about 1240 × 1754 pixels, which keeps ordinary text crisp.
Limits and tips
- Image presets cannot be undone. Always keep your original file.
- Compressing an already-compressed file again rarely helps and can lower quality further.
- For web publishing, the PDF Optimizer linearises a file so the first page appears while the rest downloads.
- If you need searchable text after an image preset, run OCR PDF on the result.
- Removing pages you do not need with the PDF Page Deleter is often the quickest way to a smaller file.
Frequently asked questions
Which preset should I start with?
Does Lossless reduce quality?
What do Balanced, Strong and Extreme change?
I need the file under a specific size. Is this the right tool?
Why did Strong return a file that is not smaller?
Is my PDF uploaded to compress it?
How do we know nothing is uploaded? Test it yourself on the privacy proof page.