Compress PDF

Try a non-rasterizing PDF rewrite. Optionally enable image-based compression, which replaces document text and structure with page images. Savings depend on the file.

πŸ—œοΈ

Drop a PDF file here or click to browse

How PDF Compression Works

By default, only the non-rasterizing PDF rewrite runs. The image-based attempt runs only when you explicitly allow rasterization for the loaded file; then the smaller successful result is selected. The result message identifies the method. Size alone does not establish that text or accessibility structure was preserved.

The rasterization trade-off

Image-based compression does not re-compress the pictures inside your PDF. It photographs the entire page. Everything that made the file a document rather than a picture is gone from the output: no selectable text, no search, no copy-paste, no screen-reader access, no working links. The output contains page images rather than the original text layer.

File size cannot safely identify text documents: an image-heavy PDF may contain an OCR text layer, links, or accessibility tags even when rasterization produces fewer bytes. Rasterization is off by default and consent resets when another file is loaded. Leave it off when searchable text or document structure matters.

Rasterization can reduce image-heavy files, but scans may also contain hidden OCR text. Enable it only if you accept losing searchable text, links, forms, bookmarks, tags, and signature validity in the image-based output.

Quality and DPI

Both settings only affect the optional image-based attempt after you explicitly enable rasterization; they do not affect the default non-rasterizing rewrite. Quality is the JPEG quality factor, and DPI controls how many pixels each page is rendered at β€” 150 DPI gives about 2.08 times the linear resolution of the 72 DPI baseline, so it produces about 4.34 times the pixels.

DPI is clamped between 72 and 300 and defaults to 150. Values outside that range are pulled back to the nearest end rather than rejected.

Tips for Best Results

Limitations

Frequently Asked Questions

Everything runs in your browser. A non-rasterizing rewrite is attempted by default. Only with your explicit rasterization consent does the tool also render pages to JPEGs and select the smaller result. Image-based output loses searchable text, links, forms, bookmarks, and accessibility tags. Neither method promises signature or complete document-structure preservation; keep the original and review the result.
Yes. Everything happens 100% client-side in your browser. Your PDF is never uploaded to any server. You can confirm this by opening your browser's Network tab β€” no file data is transmitted. Local processing avoids uploading the PDF, but browser extensions, device access, and downloaded copies still affect confidentiality.
The file guard is 100 MiB (104,857,600 bytes). Large page counts or complex pages can exhaust memory below that limit. The optional image-based pass renders pages to Canvas, so timing and responsiveness depend on the document and device.
It depends far more on what the file contains than on any setting. Scanned documents, photo albums and brochures may shrink, depending on their existing encoding and the chosen settings. A text-heavy PDF may not shrink at all, because there is no bulky image data to squeeze and the text is already compressed; on an already-optimised file the rewrite can even come out a few bytes larger. When nothing beats the original, the tool says "No reduction" and gives you no download rather than handing back a bigger file.
The optional image-based route uses lossy JPEG encoding and removes searchable text, links, forms, bookmarks, and accessibility tags. The default rewrite avoids rasterization but can still change document-level structures and invalidate signatures. Inspect the downloaded PDF rather than assuming that the smaller output preserves every feature.
Leave Allow rasterization unchecked so pages are not converted to images. Keep your original and test selection, search, links, and accessibility in the rewritten output. Scans may contain OCR text; enabling rasterization can destroy that text even when the resulting file is smaller.