pdfpix

Compress PDF without losing quality

or drop one here
Compression level
Target size
Choose a PDF file first.

Quality here has a definition, not a feeling. Two of the four compression methods never touch a pixel: one subsets fonts to the glyphs the pages draw, one rebuilds the document and drops orphaned objects. Both leave the page identical at any zoom. This page runs those first and tells you when they were enough.

How to compress a PDF without losing quality

  1. Choose the PDF

    Press Choose a PDF, or drop the file on the page. The file is read in this tab and is not sent anywhere.

  2. Name the size

    Choose a compression level, a target below the file's current size, or type your own smaller number in KB.

  3. Press Compress

    Four methods race: copy the pages into a fresh file, re-encode the placed images, cut each embedded font down to the glyphs the pages draw, and, only when no other method reached the target, redraw the pages as images with the text kept on top.

  4. Read the check, then download

    Every candidate is opened again and its characters are counted against the source. The page shows the size, the method that won and the characters kept before it offers the download.

The four methods are described in full, with the measurements behind them, in how it works.

What lossless actually means in a PDF

A PDF carries three kinds of weight. Fonts, which ship with every glyph they were designed with while the page draws a few dozen. Dead structure, meaning orphaned objects and the old revisions a file keeps every time somebody saves it. And images, which are the only part where making the file smaller can make the page worse. Only the third one forces a choice: you reduce the image data, or you keep the pixels and accept the size.

Work on the first two and the rendered page is byte for byte the same. That is what this target does first. It is not a quality slider set high, it is a different operation that has no quality to set.

The two things this page measures

Text retention is checked on every candidate: the output is re-opened, its characters are read back and compared with the source, and anything below 98% of the non-space characters is refused. On the published run of 120 cases, 102 documents carried text and every one came back at 100%, with zero candidates refused.

Image resolution is the second. When the lossless methods are enough, no image is re-encoded at all and the original pixels ship unchanged. When they are not, the compress tool says which method won, so you always know whether a pixel was touched.

When lossless is not enough

Sometimes the honest answer is that the file cannot get much smaller without a visible cost. The least compressible document on the public run still gave back 3.8% of its size, and that is a real result rather than a failure. The median across the whole run was 59.2%.

If you have a hard cap to meet, pick the number instead of the principle. A 1 MB target or the large file page will get you there and tell you exactly what it cost.

Questions

Can you compress a PDF without losing quality at all?
Yes, when the weight is in fonts and dead structure rather than in images. Subsetting fonts and rebuilding the document change nothing on the rendered page. Whether it is enough depends on the document, and the tool reports which method won.
How much smaller does lossless compression get a file?
It varies enormously. On the published run the median cut across all methods was 59.2% and the smallest was 3.8%. A font heavy text export can lose most of its weight losslessly. A scan cannot, because its weight is pixels.
How do you prove the quality was kept?
Every candidate is re-opened and read back before it ships. Page count is compared, characters are compared, and a candidate below 98% retention is discarded. The result also names the winning method, which tells you whether any image was re-encoded.
Is lossless compression slower?
No, and it is often faster, because the methods that do not touch images finish sooner and stop the race. On the published bench the median run was about 2.5 seconds including the read back check.

The other pdfpix tools

All 30 of them run in a browser tab like this one. Each reads its own output back and checks it before offering the download, and none of them needs an account.

Where to go next