PDF compression explained

The ToolAcre PDF Toolkit does not compress PDFs. That is a deliberate omission, not a missing feature: the technique available to a browser rasterises every page, which destroys selectable text and frequently makes a text document larger rather than smaller.

Almost all excess size in a PDF is embedded images. Reducing those before the PDF is built, or rebuilding the document from a smaller source, gets a real saving without damaging the file. This page explains how to find the weight and what actually helps.

Last reviewed: 2026-09-19. Checked against the tool’s own source on that date.

What makes a PDF large

A PDF is a container. Its text is stored as instructions plus embedded font subsets and is remarkably compact — a hundred pages of prose is often well under a megabyte. Its images are stored as encoded streams and are almost always where the size is.

The usual culprit is a scan or a photograph dropped in at full camera resolution. A phone photo can be four thousand pixels wide; printed into an A4 page it is displayed at a fraction of that, and the extra pixels are carried in the file forever, doing nothing.

Embedded fonts matter at the margins. A document embedding several full font families rather than subsets can carry a few hundred kilobytes of typeface. Annoying, rarely decisive.

The three things "compress PDF" can mean

Rasterising is what many browser-based "PDF compressors" do, because a browser can already render a page to a canvas and write images into a new document. It reliably produces a smaller file for a scanned document. For a text document it destroys the text layer — no selection, no search, no screen-reader access — and the resulting page images are often bigger than the vector instructions they replaced.

A tool that silently swaps your searchable contract for a stack of pictures has taken something from you that you did not agree to lose. Rather than ship that behind a friendly "Compress" button, this toolkit does not offer it and says why.

  • Downsampling images: re-encoding embedded pictures at a lower resolution or quality. This is where nearly all honest savings come from, and it is lossy for the images but harmless to the text.
  • Stream compression: applying or improving the deflate compression on content streams. Most writers already do this, so the saving is usually small.
  • Rasterising: rendering each page to an image and rebuilding the PDF from those pictures. This is the technique easily available in a browser, and it is the destructive one.

What to do instead

Work out where the weight is before doing anything. If the page count is high and the file is small per page, there is nothing to win. If a ten-page document is 40 MB, it contains images, and those images are the whole problem.

Shrink the images before the PDF exists. If you are building the document from photographs or scans, resize and re-encode them first — a 1600-pixel-wide JPEG at quality 80 is plenty for a full-width picture on an A4 page, and is a small fraction of the original camera file. Then assemble the PDF from those.

If the PDF already exists and you do not have the sources, the honest answer is that you need a tool that can rewrite image streams in place — a desktop PDF editor or a command-line tool such as Ghostscript or qpdf. That is outside what a browser can do without rasterising, so this site points you at it rather than faking it.

Two cheap wins are worth checking first. Export again from the original application with its "smaller file size" or "web" preset, which usually downsamples images properly. And if the document is a scan you only need to read, converting it once at a sensible resolution beats any post-hoc compression.

Worked example: a 46 MB "brochure" that should be 3 MB

A six-page brochure exports at 46 MB. Six pages of layout cannot account for that, so the weight is images. The source file contains six full-resolution photographs, each about 4000 by 3000 pixels straight off a camera.

On an A4 page a full-bleed photograph is about 8.3 inches wide. At 150 dpi — ample for anything that is not a fine-art print — that is roughly 1250 pixels. The photographs are carrying more than three times the linear resolution the page can show, which is about ten times the pixel count.

  1. Resize each photograph to 1500 pixels on the long edge.
  2. Export each as JPEG at quality 80.
  3. Rebuild the brochure from the resized images.

Result: Six images of roughly 300-450 KB each instead of 6-9 MB each, and a PDF in the low single-digit megabytes. The text layer is untouched and still selectable, because nothing was rasterised.

Open the tool

Shrink the images before you build the PDF

The Image Converter resizes and re-encodes JPEG, PNG and WebP in your browser. It does not open PDF files.

What this does not cover

  • This page explains the problem; the ToolAcre PDF Toolkit has no compression feature and will not gain one by rasterising.
  • The ToolAcre Image Converter can shrink the images you feed into a PDF, but it cannot reach inside an existing PDF to replace them.
  • Ghostscript and qpdf are named here as accurate answers, not as products of this site. Nobody here maintains them and no relationship exists.
  • There is no OCR anywhere in this toolkit, so a scanned page cannot be turned back into searchable text.