Compress a PDF in your browser by rebuilding pages as JPEG images at a chosen DPI — a lossy method, labelled as such, with exact byte counts. Never uploaded. Runs entirely in your browser — your numbers never leave this device.
How this works: every page is rendered to an image and rebuilt into a new PDF. That is a lossy method — the rebuilt file has no selectable text layer, and links and annotations do not survive. It shrinks scan-heavy PDFs well; born-digital text PDFs may not shrink at all.
No PDF yet. Choose one above, pick a DPI and quality, then compress.
| Preset (DPI) | A4 pixels | Quality | Typical use |
|---|---|---|---|
| 100 DPI | 827 × 1169 | 75 | Smallest output; email attachments read on screen |
| 150 DPI | 1240 × 1754 | 75 | The balanced default for scans and photo PDFs |
| 200 DPI | 1654 × 2339 | 75 | Larger scans that will be zoomed or printed casually |
| 300 DPI | 2480 × 3508 | 75 | Rarely smaller than the original — only for very heavy scans |
Your file is processed entirely in your browser and never uploaded: it is opened on your own device, each page is rendered to pixels at the DPI you choose, encoded as a JPEG at the quality you choose, and a new PDF is built holding one image per page, then offered back as a download. Nothing reaches a server, and the tool works offline once loaded. This is the one tool in this category that re-renders rather than copies, and everything below follows from that fact, stated without decoration.
The method is lossy, and labelled so. Rebuilding a page as a JPEG discards the original page structure: the text layer is removed by this process — the output's words are pictures of words, not selectable or searchable text — vector graphics become pixels, and fine detail is limited by the chosen DPI and JPEG quality. In exchange, image-heavy PDFs (scans and photo documents, whose size lives in big embedded pictures) can shrink dramatically, because a 300-DPI scan re-rendered at 150 DPI carries a quarter of the pixels. The DPI presets and their A4 pixel dimensions are in the table below; before and after byte counts are exact, measured on the real files, and the percentage is arithmetic on those two numbers.
The honest warning, which the worked examples quantify: a PDF made of vector text — the small, born-digital documents word processors produce — usually comes out larger, not smaller, under this method, because a few kilobytes of text instructions are being replaced by full-page photographs of that text. That outcome is not hidden: when the rebuilt file is bigger, the tool reports it as larger and the sensible action is to keep your original. Compress born-digital text PDFs by removing content (Delete PDF Pages) or not at all; reserve this tool for scans and photo PDFs, where it earns its name.
No — never. The file is opened, rendered, rebuilt and downloaded entirely inside your browser; it never leaves your device, is never stored or logged, and the tool works with the network disconnected.
Almost certainly because it is a born-digital text PDF. Its pages are a few kilobytes of text and vector instructions; this method replaces them with full-page JPEG photographs, which cost far more bytes. The recorded example on this page grew from 3,817 bytes to 1,024,695 bytes. The tool reports a larger result honestly rather than hiding it — for text PDFs, keep the original.
Yes, plainly. Pages are rasterised to JPEG images at your chosen DPI and quality: the text layer is removed (words become pictures of words), vectors become pixels, and detail beyond the chosen DPI is gone. That trade is worthwhile for bulky scans and photo PDFs; it is the wrong trade for documents whose text you still need to search, copy or reflow.
150 DPI at quality 75 is the balanced default — the setting behind this page's recorded results (94.1% smaller on a photo PDF, 54.3% on a scan). Drop to 100 DPI when the file is read on screen and size is urgent; go to 200 DPI when pages will be printed. At 300 DPI the rebuild is rarely smaller than the original.
Yes. Both byte counts are the real sizes of the real input and output files, and the percentage is computed from them as (1 − output ÷ input) × 100. If the output is larger, the figure is shown as a growth, in plain words, not dressed up as a saving.
More in PDF Tools · Browse all tools