How images affect PDF file size

PDF2 min read

Why one photo can outweigh a hundred pages of text, what happens when you paste an image into a document, and how to keep a PDF small at source.

In short

Text and vectors are tiny; embedded images are almost the entire file. An image is stored at its original resolution regardless of how small it appears on the page, so shrinking pictures before they go in is the most effective fix.

A hundred pages of prose in a PDF is a fraction of a megabyte. One phone photo dropped onto page three can be five times that. Understanding why makes most PDF size problems obvious.

How a PDF stores each thing

Relative cost of PDF content
ContentStored asTypical cost
TextCharacters plus a font referenceNegligible
FontsEmbedded subset of glyphs actually usedTens to hundreds of KB, once
Vector drawingsMathematical pathsNegligible
PhotographsImage data, at full resolutionAlmost the entire file

The trap: display size is not storage size

When you place a 4000-pixel-wide photo into a document and drag it down to a two-inch square, the PDF stores the original 4000-pixel image and a note saying to draw it small. Nothing is discarded. Ten such images make a 50MB document that looks like a few pages of text with small pictures.

Screenshots are the other common cause

Screenshots are usually PNG, which is lossless and very inefficient for anything photographic. A handful of full-screen PNG screenshots in a document adds up quickly. Converting them to JPEG or WebP first, at a high quality, cuts the document dramatically with no visible change.

Fixing a finished PDF

Once the images are embedded, Dexta's only option is to re-render every page as a picture — effective, but it takes your text with it. That is fine on a scan or a photo-heavy brochure and a poor trade on a text document.

Mistakes worth avoiding

  • Dragging full-resolution camera photos into a document and relying on compression afterwards.
  • Using PNG screenshots where a high-quality JPEG would be visually identical and a fraction of the size.
  • Assuming a small on-page image is a small stored image.