How to shrink a PDF without wrecking the images

A 40 MB PDF is rarely 40 MB of text. Text is tiny. A page of dense prose costs a few kilobytes. When a PDF is large, something inside it is a photograph, and that photograph is almost certainly bigger than it needs to be.

Find out what is heavy before you compress

Compressing blindly is how documents get ruined. A slide deck with screenshots and a scanned contract have opposite needs, and the same compression setting will do the wrong thing to one of them. Open the file and look at what is on the pages first.

• Photographs and screenshots: the usual culprit, and the place to spend your effort
• Scanned pages: every page is an image, even the ones that look like text
• Embedded fonts: a few hundred kilobytes, worth leaving alone
• Actual text and vectors: effectively free, never the problem

Resolution is the lever that matters

A camera photo dropped into a document arrives at something like 4000 pixels wide. Printed at letter size that is roughly 500 DPI. Print shops ask for 300 DPI, and a screen shows about 96 DPI more or less. So the file is carrying four or five times the pixels anyone will ever see.

Downsampling images to 150 DPI is the single change that does the most work. A document going out by email routinely drops from 40 MB to under 4 MB, and on screen it looks the same, because those pixels were never being displayed.

150 DPI for anything read on a screen. 300 DPI if it will be printed. Above 300 you are storing detail that no output device will reproduce.

Rule of thumb

Where it goes wrong

Scanned text is the exception that catches people out. A scan is an image of letters, and letter edges are exactly the high contrast detail that JPEG compression handles worst. Push a scanned contract too hard and the text turns muddy and soft, which is fine until someone needs to read a clause or run OCR over it. Keep scans at 300 DPI and accept a larger file.

The other trap is compressing a file that has already been compressed. Each pass re-encodes the images and adds artifacts to artifacts. If a PDF looks blocky already, going again will not shrink it much and will make it worse. Go back to whatever produced it and export once, properly.