PDF Compress — Reduce PDF File Size in Your Browser
Compress PDF files by re-encoding their embedded images. Three quality levels, runs entirely in your browser. No uploads, no accounts, no limits.
Drop your PDF file here
or click to browse (max 100MB)
PDF compression is a two-axis trade-off — image quality on one side, file size on the other — and the third axis nobody talks about: what the PDF is for. A PDF that is going to be printed at 300 dpi needs different compression than one that will be read on a phone, and a PDF that will be signed or re-OCR’d later needs the text layer preserved. This tool is built for the first two cases: get the file small enough to email, embed in a slide deck, or upload to a CMS, without paying for a server-side service.
The honest way to compress a PDF is to look at what is taking up the bytes. For a typical 10-page document, the breakdown is usually about 60% embedded images, 30% font programs and content streams, and 10% structural overhead. Re-encoding images is where the saving lives — fonts and content streams are already as small as they can reasonably get without breaking the text layer. The three presets here expose the same trade-off that Ghostscript’s /screen, /ebook, and /printer settings make on the command line, with per-image quality sliders instead of a single switch.
If the goal is sharing a one-time view, aggressive compression is fine. If the PDF is going to be printed, archived, or used as the source for further editing, stay on Light or use a different tool. For scanned PDFs with no text layer, run an OCR pass first and then compress — the text layer makes the file searchable and the OCR engine usually downscales images anyway, so the combined saving is bigger than compressing the raw scan.
How to use
Drop a PDF
Drag a .pdf file into the drop zone, or click to browse. Up to 100 MB; the file is read locally and never leaves your device.
Pick a compression level
Light preserves quality (re-encodes JPEGs at 85%). Medium downscales images to 85% and re-encodes at 70% — recommended for most documents. Aggressive hits 70% scale and 50% quality for the smallest output.
Download the compressed file
The compressed PDF is rebuilt from the original's text, images, and layout. Saved bytes, original size, and re-encoded image count are shown before the download button.
Frequently asked
What does this tool actually do?
It walks the PDF page by page, extracts every embedded image, re-encodes them at a lower JPEG quality, and rebuilds the PDF with the smaller images. The text layer is preserved; vector shapes are flattened to images during the rebuild.
Will the output look identical to the input?
No. This is a lossy compressor — it trades image quality for size. At Light, the difference is usually invisible at screen zoom; at Aggressive, expect visible banding and softening in photos.
Why didn't my PDF get smaller?
If the source PDF is mostly text or already uses highly compressed images, the saving will be small. Text-only PDFs only benefit from object-stream repacking, which usually yields under 5%.
Are form fields, annotations, and layers preserved?
No. The rebuild is a flat rendering — form fields, JavaScript actions, embedded fonts, and annotations are stripped. For those, use a tool that operates on the original PDF stream instead.
Is there a file-size limit?
The drop zone caps uploads at 100 MB. The actual limit is your browser tab's available memory — a 50 MB PDF with 200 images can use 500 MB of RAM during rebuild.
Limitations
- Lossy outputEvery preset re-encodes images as JPEG at a lower quality. Photos soften; screenshots with text may develop JPEG artifacts around the letters.
- Vector graphics are flattenedCharts, CAD drawings, and any vector content are rasterised into the page output. The file gets smaller but is no longer resolution-independent.
- No form fields or annotationsForm fields, JavaScript actions, embedded fonts, layers, and digital signatures are dropped during the rebuild. The output is a flat document.
- Not a substitute for OCRIf the source is a scanned PDF without a text layer, the text in the output is the same scanned image. Run an OCR pass first if you need searchable text.
Platform notes
- macOS
- Works in Safari 16+, Chrome, and Firefox. macOS Preview can re-save PDFs with similar image compression, but this tool exposes the per-image JPEG quality so you can tune the trade-off.
- Windows
- Edge and Chrome both support the canvas-based image re-encoding. Acrobat Pro has a 'Reduce File Size' feature that does the same job on a desktop install.
- Linux
- Runs in any modern browser. For command-line work, Ghostscript with -dPDFSETTINGS=/ebook gives a similar result, but cannot be tuned per-image the way this tool can.
- iOS
- Safari on iOS 16+ supports the re-encoding pipeline. On older iOS, the canvas API rejects large images (>~4000px on the long side) and re-encoding falls back to the original.