How to compress a PDF
Drop a PDF onto the page. Choose one of two modes:
Smart compression (default, recommended)
Preserves text, vector graphics, and selectable words. Only images embedded inside the PDF are re-encoded at lower quality. Metadata and embedded thumbnails are stripped. This is how Adobe Acrobat and iLovePDF achieve real size reductions on mixed-content PDFs.
Important truth: if your PDF contains no images, Smart mode can only shave off metadata (typically 5–15% savings). Text in PDFs is already stored compactly. No tool — including Acrobat — can meaningfully shrink a text-only PDF without destroying the text.
Rasterise pages
Renders every page as a JPEG at a chosen DPI and quality. Best for scanned documents and image-heavy PDFs. Text becomes non-selectable — useful if you want to prevent copying, harmful if you don't.
When to use which mode
| PDF contains | Use | Typical savings |
|---|---|---|
| Text only (report, contract, invoice) | Smart | 5–15% |
| Text + embedded images | Smart | 20–40% |
| Scanned document (image of pages) | Rasterise | 60–90% |
| Photo book, brochure, portfolio | Rasterise | 50–80% |
How target size works
In Rasterise mode, the target-size search:
- Renders every page once at maximum quality and caches it
- Tests compression combinations from highest quality to lowest
- Stops at the first attempt that lands at or below your target plus tolerance
- Never goes below your target if a larger result was available — you get the best quality that fits
So if you target 1 MB with 10% tolerance, the tool stops at the first result between 1.0 and 1.1 MB. It won't return 0.7 MB just because the previous attempt hit that.
Common questions
Why doesn't Smart mode compress text PDFs much?
Text in PDFs is already stored compactly as character codes. Without embedded images to re-encode, the only savings come from stripping metadata. Text PDFs compress typically by 5–15%, not 90%.
What does target size do?
In Rasterise mode, the tool tries different quality and DPI combinations until the output fits under your target plus tolerance, while staying as close to the target as possible from above.
Can I get exactly the target size?
Not exactly — compression is discrete, so the tool picks the closest attempt at or under target + tolerance. That's the honest answer for any raster-based compressor.
Will text stay selectable?
Yes in Smart mode. No in Rasterise mode.
Are my PDFs uploaded?
No. Everything happens in your browser.