Compress PDF

What the tool actually does to your file

This is the honest version, because it changes when you should and should not use it. The PDF is opened in your browser with pdf.js. Each page is redrawn onto a canvas at a chosen resolution, that canvas is saved as a JPEG at a chosen quality, and the JPEGs are assembled into a brand new PDF by pdf-lib. The original file is never uploaded. Everything, including the rebuilt document, is produced on your own device, and the finished file downloads straight from the tab.

Rebuilding pages as images is the one compression method a browser can perform reliably without a server, and it is very effective on documents that were already pictures. It is the wrong tool for some files, which is covered below.

The trade off you are accepting

Because every page becomes an image, the text on the output is no longer text. It cannot be selected, copied or searched, a screen reader cannot read it, and no editor can reflow it. Anything that relied on the text layer, such as form fields, links and bookmarks, is gone as well. What you get in return is a file whose size is controlled entirely by image settings rather than by whatever the original producer embedded.

So the decision is simple. If the PDF is a scan, a photographed document, a slide export, or anything you only need people to look at, compressing this way is a good fit. If it is a contract someone will search, a report that needs accessible text, or a form with fillable fields, do not compress it here. Reducing quality in exchange for size is a reasonable trade; losing a text layer by accident is not.

Choosing a preset

Three presets set the two variables that matter, resolution and JPEG quality.

High Quality renders at 150 DPI with JPEG quality 0.80. Pages stay crisp enough to print acceptably, and fine print and thin lines survive. Savings are the smallest of the three, and on a file that was already well optimised there may be none at all.

Balanced renders at 120 DPI with quality 0.65 and is the default. On screen it looks close to the original for ordinary text documents, with visible softening only when you zoom in. This is the right first attempt for email attachments and upload forms.

Extreme renders at 96 DPI with quality 0.50. Text edges become noticeably soft, photographs pick up visible compression artefacts, and small print can become hard to read. Use it when a hard size limit has to be met and legibility on screen is the only requirement.

Hitting a size target

Upload forms usually state a maximum, commonly a few megabytes, and the practical method is to work downwards rather than guess. Start on Balanced and check the three figures the tool reports after it runs: original size, compressed size and the percentage saved. If the result is under the limit, stop there. If it is still too large, run the original file again on Extreme rather than compressing the already compressed output, because compressing a compressed image twice degrades quality faster than going straight to the lower setting once.

If Extreme still misses the target, the document is probably too long rather than too heavy per page. Splitting it and uploading in parts, where the form allows, keeps far more quality than crushing every page. Two other things help before you compress at all: remove pages you do not need, and if the source was a phone photograph, retake or crop it so the page fills the frame instead of sitting inside a desk.

When the file does not shrink

Sometimes the compressed version comes out the same size or larger, and the status line will say so. There are three usual explanations. The PDF may be pure vector text, which is already extremely compact, so turning each page into a photograph of text adds data rather than removing it. The PDF may already contain heavily compressed images, in which case re encoding them gains nothing. Or the pages may be physically large, such as posters or plans, where even a modest DPI produces a big canvas. In all three cases the original file is the better one to keep.

Two limits are worth noting. Files over roughly one hundred megabytes trigger a warning because processing them can use a lot of memory, and on a phone a very large document may not finish at all. Encrypted documents will not open here at all, so the protection has to come off first using software that can export an unlocked version.

A worked example

A twelve page scanned tenancy agreement, photographed with a phone at full camera resolution, will not attach to a form that caps uploads at five megabytes. Open it here, leave the preset on Balanced, and compress. The pages are re rendered at 120 DPI, which is well above the resolution the original photos actually carried any detail at, so the visual result is close to identical while the file becomes a fraction of its former size. The signed pages are still perfectly readable, and because nobody needs to search the document, losing the text layer costs nothing. Download, check the first and last pages on screen, then attach.

Related tools

Use Split PDF to break a long document into parts, Merge PDF to join them again, and Images to PDF when you are starting from photos and want to control the page size from the beginning. Sign PDF handles signing before you compress, which is the right order, since signing a compressed file is fine but compressing after signing keeps the file smaller. Splitting and joining PDFs locally walks through those steps.

Questions about compression