Compress PDF
Recompress the images inside a PDF and see exactly how many bytes it saved.
Drop a PDF here
Nothing is uploaded. The file is read in this page.
Pixels. 0 keeps the original size. Downscaling large scans saves more than quality alone.
Everything here runs on your device. Nothing you enter is uploaded or stored.
Shrink a PDF by recompressing the JPEG images embedded inside it, with the quality and maximum dimension you choose. See exactly how many images were found and how many bytes were saved — not an estimate, a measured before-and-after.
How to use it
- Drop or choose the PDF you want to compress.
- The tool reports how many JPEG images it found and how much of the file’s size they account for.
- Set a quality level — lower produces a smaller file — and optionally cap the maximum image dimension.
- Press Compress to download the result, with the exact original and compressed size shown.
Why this targets JPEG images specifically
A PDF’s file size is rarely spread evenly across its content. In the overwhelming majority of oversized real-world PDFs — scanned documents, reports with embedded photographs, presentations exported with images — the bulk of the bytes live in embedded JPEG (technically, DCTDecode-filtered) image streams. Text, even a lot of it, and vector graphics are comparatively tiny in byte terms.
This tool walks the PDF’s internal object structure directly, finds every JPEG image stream, decodes it, and re-encodes it at your chosen quality (and optionally at a smaller pixel dimension), replacing the original stream only if the result is genuinely smaller. Other embedded-image encodings that occasionally appear — indexed-colour PNGs, raw uncompressed bitmaps, CMYK images — are deliberately left alone. They’re rare in practice, and modifying them correctly requires different handling than JPEG recompression; touching them for a marginal size win risks producing a corrupted file, which is a worse outcome than a slightly smaller compression ratio.
The “never larger” guarantee
Recompressing an already-efficient JPEG at a similar quality can occasionally produce a result that is the same size or even slightly larger than the original — JPEG’s compression isn’t perfectly consistent between encoders and settings. This tool checks the size of every recompressed image against its original before replacing anything, and skips any image where the new version isn’t actually smaller. That means running this tool on a PDF that’s already well-optimized will simply report that nothing could be shrunk further, rather than silently making the file worse.
What it does not do
It does not touch or optimise text, fonts, or vector graphics — only embedded JPEG raster images. It does not perform lossless recompression; lowering quality does discard image detail, the same tradeoff any JPEG re-encode makes. It cannot compress a PDF’s non-JPEG images (see above), so a document built entirely from PNG screenshots or vector diagrams may show little or no size reduction — the tool reports this honestly rather than claiming a universal fix.
Quality setting guide
| Quality | Effect | Best for |
|---|---|---|
| 80-90% | Minimal visible difference | Documents that need to look near-identical |
| 50-70% | Noticeable but acceptable softening | General file-size reduction |
| 20-40% | Visible compression artefacts | Maximum size reduction, e.g. email attachment limits |
What compresses well versus what doesn’t
| Content type | Typical result |
|---|---|
| Scanned document pages (JPEG) | Significant reduction |
| Embedded photographs (JPEG) | Significant reduction |
| Screenshots or diagrams (often PNG) | Little to no reduction |
| Text-only pages | No reduction — nothing to compress |
Questions
Why does this only compress JPEG images?
Scanned pages and embedded photographs — the actual source of most oversized PDFs — are stored almost universally as DCTDecode (JPEG) streams inside the PDF. Other embedded-image encodings (indexed colour, CMYK, raw FlateDecode) are a small minority of bytes in most real documents, and rewriting those risks introducing a corrupt file for a small size gain. This tool is honest that it targets JPEG content specifically rather than implying it touches everything.
What if my PDF has no images to compress?
The tool tells you before you commit — if it finds no JPEG images, it says so directly rather than producing a "compressed" file that is actually the same size or larger. Text-heavy and vector-graphic PDFs typically have little to gain from image recompression.
Will compressing reduce text quality or make it blurry?
No. Text and vector graphics in a PDF are not raster images and are not touched by this process at all — only embedded JPEG image data is recompressed. Text stays exactly as sharp as the original at any zoom level.
Is a re-encoded page ever allowed to come out larger than before?
No. Each image is only replaced if the recompressed version is genuinely smaller than the original; if recompression would make an image larger, that image is left untouched. This means the tool can safely be used speculatively without risk of the file growing.
Last updated