Compress PDF — Reduce File Size Without Uploading

Compress large scanned PDFs entirely on your device — no upload, no size cap.

PDF Compressor targets the one thing that actually bloats a scanned document: the raster images inside it. It decodes each embedded image, decides whether recompressing it as JPEG is the right call, resizes it to match a target DPI, then deduplicates repeated font programs and drops stale metadata — all inside a Web Worker so a 200-page scan never freezes the tab. Nothing is uploaded, and the result is never allowed to end up bigger than what you started with.

How to compress a PDF

  1. Drop a PDF onto the Load PDF panel, or click it to pick one; a preview of the document appears underneath.
  2. Leave the preset on Recommended (150 DPI, quality 60), or switch to Custom and set your own target DPI (72–300) and JPEG quality (20–95).
  3. Click Compress PDF. Reading, image recompression, font cleanup and saving all run in a background worker inside the tab while a progress bar names the current stage.
  4. Compare the original and compressed sizes, then click Download compressed PDF. If the new file would not have been smaller, you get your original bytes back instead.

Which images actually get recompressed

Every raster inside a PDF page is its own indirect image object, and each one gets classified before anything happens to it. Images already encoded as JPEG are the primary target, since decoding and re-encoding them at a lower quality is where most of a scan's size actually lives. Uncompressed or Flate-encoded 8-bit grayscale and RGB images go through a second check: a sample of their pixels, quantized into 16-step color buckets, decides whether the content reads as photographic or as flat line art.

Anything that lands on the flat side of that split — scanned black-and-white text, logos, simple diagrams — is left byte-for-byte untouched, because it's already stored losslessly and JPEG would only make it blurrier for no size benefit. Indexed-color images, sub-8-bit images, CMYK images, and any image carrying an alpha mask are skipped for the same reason: none of them have a safe, lossless path through a JPEG re-encoder, and JPEG has no alpha channel to preserve one in anyway.

What the DPI and quality sliders control

The DPI slider (72 to 300, defaulting to 150) isn't a fixed pixel target — it's converted per page, using that page's own physical size in points, into a maximum pixel footprint an image is allowed to occupy. An image already smaller than that footprint is never upscaled; only genuinely oversized scans get shrunk, and only down to the resolution the page can actually display at your chosen DPI.

The quality slider (20 to 95, defaulting to 60) is the MozJPEG re-encode quality applied to whichever images passed the photographic check above. Sixty is tuned for a typical scanned contract or tax form: legible at 150 DPI without visible blocking on the kind of photographic-but-not-glossy content most scans contain.

Fonts and metadata get cleaned up too

Font programs embedded more than once — common when the same typeface shows up across many pages of a scanned form — are hashed by content and collapsed to a single copy, with every page's font reference repointed at the first one found. The document's XMP metadata stream, a frequent source of stale, bulky editing history, is dropped outright, and the Producer field is reset to something honest. Title and Author are left alone deliberately, since they're user-meaningful rather than junk — a narrower cleanup than PDF Stamp's own metadata-strip mode, which clears every field on request.

Why the file you download is never bigger than what you uploaded

Two separate safety nets protect against making things worse. Per image, if a JPEG re-encode comes back larger than the stream it would replace, that image keeps its original bytes rather than upgrading to a bigger file. At the very end, the whole saved PDF is compared against the original upload one more time — structural overhead like a rewritten cross-reference table can occasionally erase a small per-image saving — and if the total somehow crept past the original size, you get the untouched original back instead.

The compressor works on one PDF per run. On the free tier it sets no page limit, no file-size limit and no daily count for this tool, so the real ceiling is how much memory your device can give the tab.

Your PDF never leaves your device during compression

Most online compressors ask you to upload the document, process it on their server and promise to delete it a few hours later. Here there is no server step at all: the file is read into the page, handed to a Web Worker, recompressed and saved back to a download link without a single request carrying its bytes. With no server copy, there is nothing to trust anyone to delete.

You can check this yourself by opening your browser's developer tools on the Network tab before loading a file and watching that no request appears while it compresses.

When compression won't do much

Questions people ask about PDF Compressor

Why didn't my PDF get any smaller?
If the size barely changed, the document was probably already lossless-compressed line art or vector content rather than photographic scans — this tool only recompresses raster images it classifies as photographic, and leaves flat-color or already-optimized images untouched on purpose.
What does the Recommended preset actually set?
Recommended applies 150 DPI and MozJPEG quality 60, a pairing tuned for a typical scanned document: legible text at normal reading zoom without the blocky artifacts a lower quality setting would introduce on photographic-looking scan noise.
Does compressing a PDF reduce quality?
Only in the photographic images it chooses to re-encode, which are saved as JPEG at your chosen quality and scaled down if they exceed the target DPI. Text, vector drawings and flat line-art scans are never re-encoded, so they stay exactly as sharp as before.
Can I compress a PDF without losing quality?
Not fully: the font deduplication and metadata cleanup are lossless, but the image step is JPEG and therefore lossy. To keep visible change to a minimum, choose Custom and raise the DPI toward 300 and the quality toward 95; you will save less space in exchange.
How much smaller will my PDF get?
It depends on what the file contains. Photo-heavy scans usually shrink the most, while text-only or vector documents barely move. After each run the tool shows the original size, the compressed size and the percentage saved, and it never hands you a larger file.
Is it safe to compress confidential PDFs here?
The document is processed inside your own browser tab by a Web Worker, and no network request carries its contents, so there is no server copy of a contract or medical record to worry about. Closing the tab discards the loaded file and the result.
Can I compress a password-protected PDF?
No. The compressor stops with an error on any encrypted PDF, including files that open without a password but restrict printing or copying. Remove the protection first with PDF Password Remover, which needs the current password, and then compress the unlocked copy.
Will compressing remove my document's title or author?
No. The metadata cleanup this tool runs keeps Title and Author intact and only removes the bulky XMP metadata stream and resets the Producer field; if you want every metadata field wiped, PDF Stamp's dedicated strip mode does that instead.
Is there a limit on how many PDFs I can compress?
Not on the free tier. You compress one document per run, but there is no daily count, no page ceiling and no file-size cap for this tool, so you can run it again and again; only very large files may hit your device's memory.

Related tools on BrakeTools

Content reviewed 2026-09-27.