Home › PDF tools › Compress PDF

Compress PDF

Reduce a PDF's file size in your browser with two clearly labelled modes: Lossless, which re-saves the file and keeps text selectable, and Strong, which rasterises every page to JPEG. You get the original size, the output size and the percentage saved — measured on your file. Nothing is uploaded.

Choose a PDF file

FileNo file loaded

Pages–

Original size–

Compressed size–

Saved–

No PDF loaded yet.

Compression mode

Lossless — keeps text selectableRe-saves the PDF with object streams and removes the document metadata. No page is redrawn: text stays vector and selectable, images keep their original pixels. This only helps files that carry bloat — many PDFs will barely change, and the readout will tell you. Strong — rasterises pagesDraws every page as an image and re-encodes it as JPEG. Text stops being text: it can no longer be selected, copied or searched, and links and form fields are lost.

Lossless mode re-saves the file and strips metadata. Text stays selectable and quality is untouched, but most files change only a little.

Choose a PDF file to begin.

The sizes and the percentage shown after a run are counted from your own file: the tool reads the bytes you loaded and the bytes it wrote, then divides them. Nothing here is an estimate, and the page makes no promise about how much your file can lose. If a run does not make the file smaller, you are told so and no download is created.

What this PDF compressor does

This page reduces the size of a PDF in your browser. You pick a file, choose one of two clearly labelled modes, press Compress PDF, and the tool reports the exact original size, the exact output size and the percentage that was saved — counted on your document, not copied from anybody else's. Then you download the new file. Your original PDF on disk is never modified or overwritten.

The two modes do genuinely different things, which is why they are named rather than presented as one slider. Lossless re-saves the document with object streams instead of plain cross-reference tables and drops the document information dictionary and any XMP metadata. Not a single page is redrawn, so text stays vector text and images keep their original pixels; the gain comes only from overhead the file was carrying. Strong renders each page with the pdf.js renderer at 72, 100 or 150 DPI, encodes that render as a JPEG at the quality you set, and rebuilds the document from those images. Page dimensions are preserved, but the page content is now a picture. The warning sits next to the DPI and quality controls, because that is the moment the trade-off matters.

Nothing is uploaded. The file is read from your disk with the browser's file API, processed inside the page by pdf-lib and pdf.js, and handed back to you as a download. There is no server, no account, no queue and no copy of your document anywhere except your own device. Once the page has loaded you can disconnect from the network and both modes still work.

How to use this PDF compressor

Press Choose a PDF file and pick the document. The page shows the file name, the page count and the original size, and nothing leaves your browser.
Choose a mode. Lossless keeps text selectable and only removes overhead; Strong redraws each page as a JPEG image, and the warning beside those controls spells out what that costs.
If you chose Strong, set the render resolution (72, 100 or 150 DPI) and drag the JPEG quality slider. In Lossless mode there is nothing to tune, because there is no quality decision to make.
Press Compress PDF. Strong mode works through the document one page at a time and reports its progress as it goes.
Read the readout: original size, compressed size and the percentage saved, all measured from the file you loaded. If the run produced something bigger, the page says so and offers no download.
Press Download to save the result under a new name, for example report-compressed.pdf or report-strong.pdf. Your original stays exactly as it was.

Worked example

The test files used here were built for this page. The first is site-report.pdf: 6 A4 pages of text, 13,693 bytes (13.37 KB), carrying a full set of document metadata. Every number below is what this page reported in a browser on that file.

Lossless — keeps text selectable: 13,693 bytes in, 7,066 bytes out (6.90 KB) — 48.4% smaller, 6,627 bytes saved. All 6 pages are still there and the text is still text: extracting the text from the downloaded file returns the same 6 pages and the same 18,984 characters as the original. About 390 bytes of that saving came from removing the eight metadata fields; the rest came from writing the object streams and cross-reference streams instead of plain tables.
Strong — rasterises pages, 100 DPI at JPEG quality 0.70: the same file came out at 875.72 KB, which is 862.35 KB larger than the original, so the tool offered no download and told you why. That result is worth seeing once: rasterising a file that is mostly vector text usually makes it bigger, because a 100 DPI JPEG of an A4 page is heavy on its own — this run averaged roughly 146 KB per page.

The second test file is the case rasterising is actually for. scan-bundle.pdf is 3 A4 pages totalling 310.53 KB, and each page is one large JPEG — what a scanned bundle looks like:

Lossless on the scan: 310.53 KB to 310.34 KB — 0.1% smaller, 193 bytes. There was nothing to remove: the file already used object streams, and a scan carries no font or resource overhead worth mentioning, only the images.
Strong on the scan, 100 DPI at quality 0.70: 310.53 KB to 94.43 KB — 69.6% smaller. The downloaded file was exactly 96,696 bytes. Nothing was lost in kind, because those pages were images to begin with; what changed is their resolution and JPEG quality.

One more control, because it is the honest counterpart to the headline number: the same lossless mode on a 6-page file that was already written with object streams (7,408 bytes, 7.23 KB) returned 7,066 bytes — a 4.6% saving, 342 bytes, all of it metadata. Two files of the same page count, and lossless had almost nothing left to do once the structural overhead was already gone. That is exactly why no percentage is printed in advance anywhere on this page: the number depends on the file, and the tool measures it during the run.

Who is this PDF compressor for?

Anyone fighting an upload or attachment limit: application portals, email gateways and form systems reject files over a fixed size. You load the file, look at the measured saving, and decide whether the lossless result is enough or whether the text has to be traded for a raster image.
Office and admin work: shrink a bulky export before it goes into a records system or a shared drive, without installing a desktop editor and without sending an internal document to an unknown server.
Students and researchers: make a lecture pack or a submitted assignment smaller, and use the lossless mode when the document still has to be readable by a search tool or a marking system.
People handling confidential papers: contracts, payslips, medical letters and HR files can be compressed on a work laptop because the file is never uploaded, never logged and never handed to a third party.
Scanned-document work: a scan is already an image, so Strong mode at a sensible DPI and quality is the mode that actually moves the number, and the page shows you the measured result rather than a slogan.

Why browser-side compression has limits

A desktop command-line tool such as Ghostscript can do something a browser page cannot easily do here: it opens the internals of the PDF, re-encodes the embedded images at a lower resolution or as JPEG, subsets fonts so only the used glyphs remain, drops unused objects, and converts colour spaces that cost more bytes than they are worth. The two libraries this page uses do not include that pipeline. pdf-lib can restructure a file — object streams, cross-reference streams, metadata — but it does not recompress the images inside it, because re-encoding an image is exactly the operation that destroys information. pdf.js can render a page faithfully, but the only way to write that render back into a document from the browser is to store it as a picture.

That leaves the two honest options on this page: restructure the file without touching a page, or render the pages and accept that they become images. There is no middle setting that quietly recompresses your photos at 60 percent and keeps your text, because nothing in the browser can do that without a full PDF image pipeline.

Lossless gains are naturally modest. Metadata is usually a few hundred bytes; object streams and a compressed cross-reference table save what the plain tables cost. If your file came from an application that already writes object streams — most modern PDF writers do — the re-save can come out the same size, or even a little larger, and this page will say so instead of handing you a pointless download. A file that carries no bloat has nothing to lose.

Rasterising trades text for size, and it has a floor. At 150 DPI an A4 page is roughly 1,240 by 1,754 pixels, and a JPEG of that size cannot go below a certain number of bytes no matter how low the quality setting goes. So the smallest file you can reach is governed by the page count, the DPI and the JPEG quality — not by anything this page can promise. And the trade cannot be undone afterwards: the output is a set of images, and no later tool can turn it back into a searchable document.

This is why nothing on this page says "up to 90% smaller", or any other figure. Every number you see after a run was produced by that run, on that file, a second earlier. If you need the compression that only a full image-recompression pipeline gives, a desktop tool is still the right answer for that job; what this page offers is the part of the job a browser can honestly do, done privately.

	Lossless — keeps text selectable	Strong — rasterises pages
What happens to a page	Nothing. It is copied as it is.	It is drawn, then stored as a JPEG image.
Text	Still text: selectable, copyable, searchable.	Pixels. No selection, no search, no screen-reader text.
Images inside the file	Original pixels, original resolution.	Replaced by one render per page, at your DPI and quality.
Links and form fields	Kept.	Lost.
What sets the saving	How much metadata and structural overhead the file carried.	Page count, DPI and JPEG quality.
When it is the right choice	Any file that must stay a document you can read, quote or search.	A file that only has to look right — a scan, a printer proof, a preview.

Both columns describe what the code on this page does. Nothing in the table is a claim about results: the sizes after a run are printed next to the sizes before it, and that is the only measurement the page offers.

Common use cases

Getting a document under an email attachment limit before you send it, and knowing exactly how many bytes you saved.
Uploading a PDF to a portal that caps file size, where the readout tells you whether you are under the limit before you try again.
Stripping the author, title and software fingerprints from a PDF before you send it to someone outside your organisation.
Turning a scanned bundle into a smaller file for sharing, at a DPI and quality you choose rather than one a website chose for you.
Bundling several page images into one shareable PDF for a portfolio, an insurance claim or a proof for a printer.
Keeping a copy of a large report that opens quickly on a phone, while the full-quality original stays on your disk.

Pro tips

Always try Lossless first. It costs nothing in quality, so the only thing you can lose is a few seconds. If the readout shows a saving you can live with, stop there — the text stays text.
Watch the readout, not the mode name. A lossless run that saves almost nothing means the file had no structural bloat; that is a result, not a failure.
On scans, compare 100 DPI and 150 DPI. Run the tool twice and compare the two measured sizes. If 100 DPI is readable for your purpose, the difference in bytes is usually worth having.
Raise the quality before you raise the DPI. JPEG quality affects how the text edges look; DPI affects how much detail survives. If small print smears, quality is usually the first control to move up.
Work in a desktop browser for long documents. Strong mode holds a canvas and a page image in memory as it goes, and large files plus many open tabs are the usual reason a run feels slow.
Keep both copies when it matters. Download the compressed file under a new name and leave the original untouched, so a rasterised version never becomes the only copy you have.

FAQ

Are my PDFs uploaded to a server?

No. The file is read from your disk by your own browser and the smaller file is written back to your device by pdf-lib. Nothing is uploaded, no copy is stored on a server, and no account or email address is needed.

Which compression mode should I choose?

Start with Lossless if the text has to stay selectable, copyable or searchable — it cannot damage the page content. Switch to Strong only when you need a substantially smaller file and can accept that every page becomes a JPEG image.

Does Lossless mode change anything on the pages?

No. Lossless mode does not redraw, re-compress or re-sample a single page. Text stays vector text, images keep their original pixels, and links and form fields stay where they were. All it changes is how the file is stored and which metadata it carries.

Why did my file barely change size?

Because the bytes were mostly page content, not overhead. Lossless mode removes metadata and rewrites the cross-reference tables as object streams, so it only helps when a file carries that kind of bloat. A PDF that was already written this way can lose almost nothing. In that case use Strong mode, which redraws each page as an image.

What exactly does Lossless mode remove?

The document information dictionary — title, author, subject, keywords, producer, creator, creation date and modification date — and the XMP metadata stream when the file has one. It also re-saves the PDF with object streams, which replaces the plain cross-reference table with a compressed one.

Why does Strong mode sometimes shrink a file far more than Lossless?

Because it throws away the internals of each page and keeps only a picture of it. Anything invisible in the original — fonts embedded for characters the file never uses, unused resources, hidden layers, duplicate images — disappears, and the picture is stored as a single JPEG. How much you gain depends entirely on how much of the original file was that kind of overhead, so the tool reports the measured number for your file instead of promising one.

Will my text still be selectable in Strong mode?

No. Strong mode draws each page onto a canvas and re-encodes the result as a JPEG, so the output contains no text at all: no selection, no copy and paste, no search, no screen-reader text. Links and form fields are gone too, and zooming in shows the pixels of the raster image.

Can I compress a scanned PDF?

Strong mode is built for scans, because a scan is already a picture of a page. Choose a DPI that matches the detail you need and a JPEG quality you can live with, then read the reported before-and-after sizes. Lossless mode will normally do nothing useful for a scan, since there is no text in it to keep and the image inside is already compressed.

Can I choose the resolution and the JPEG quality?

Yes, in Strong mode: 72, 100 or 150 DPI for the page render, and a JPEG quality slider from 0.40 to 0.95. Higher DPI keeps small print readable but stores more pixels; lower quality makes a smaller JPEG with more visible blocking around text and thin lines.

Can I compress an encrypted or password-protected PDF?

No. An encrypted file cannot be read by the tool, and the page says so instead of producing a broken download. Open the file in a PDF reader, remove the password or print it to a new PDF, then compress the unlocked copy.

Is there a page or file size limit?

There is no fixed limit, but everything is processed in your browser's memory, so very large documents are bound by the RAM and the processor of your device. Strong mode is the heavy one because every page is rendered on a canvas at the chosen DPI. For big files, use a desktop browser and close other tabs first.

Is this PDF compressor free, and does it add a watermark?

Yes, it is free, with no sign-up, no email gate and no limit on how many files you compress. Nothing is added to your document: no watermark, no stamp, no extra page. The only thing the output loses is what the mode you picked removes.