Skip to main content

Quickly check whether there is a visible difference between two PDFs, using ImageMagick and pdftocairo.

Project description

This script checks whether two PDFs are visually the same. So:

  • White text on a white background will be ignored.

  • Subtle changes in position, size, or color of text will be detected.

  • This program will ignore changes caused by a different version of the PDF generator, or by invisible changes in the source document.

This is in contrast to most other tools, which tend to extract the text stream out of a PDF, and then diff those texts. Such tools include:

There seem to be some tools similar to the one you’re looking at now, although I have experience with none of these:

The strength of this script is that it’s simple to use on the command line, and it’s easy to reuse in scripts:

from diff_pdf_visually import pdf_similar

# Returns True or False
pdf_similar("a.pdf", "b.pdf")

Or use it from the command line:

$ pip3 install --user diff-pdf-visually
$ diff-pdf-visually a.pdf b.pdf

How to install this

You can install this tool with pip3, but we need the ImageMagick and Poppler programs.

On Ubuntu Linux

  1. sudo apt update

  2. sudo apt install python3-pip imagemagick poppler-utils

  3. pip3 install --user diff-pdf-visually

  4. If this is the first time that you pip3 install --user something, then log out totally from Linux and log in again. (This is to refresh the PATH.)

  5. Run with diff-pdf-visually.

On Mac with Homebrew (untested)

  1. Run brew install poppler imagemagick.

  2. pip3 install --user diff-pdf-visually

  3. If this is the first time that you pip3 install --user something, then close your terminal and open a new one. (This is to refresh the PATH.)

  4. Run with diff-pdf-visually.

On Windows Subsystem for Linux

I’ve never tried but I think this will work. Give it a go and let me know (at bram at bram dot xyz) if it worked! Unfortunately it takes quite a while to get everything installed.

  1. Install Windows Subsystem for Linux (WSL) and Ubuntu 18.04, for instance with this tutorial

  2. Initialize Ubuntu 18.04 (tutorial)

  3. Now proceed with the Ubuntu Linux instructions.

Let me know (at bram at bram dot xyz) if this worked!

On Windows native

Lars Olafsson suggested that the following might work:

How it works

We use pdftocairo to convert both PDFs to a series of PNG images in a temporary directory. The number of pages and the dimensions of the page must be exactly the same. Then we call compare from ImageMagick to check how similar they are; if one of the pages compares different above a certain threshold, then the PDFs are reported as different, otherwise they are reported the same.

You must have ImageMagick and poppler already installed.

Call diff-pdf-visually without parameters (or run python3 -m diff_pdf_visually) to see its command line arguments. Import it as diff_pdf_visually to use its functions from Python.

There are some options that you can use either from the command line or from Python:

$ diff-pdf-visually  -h
usage: diff-pdf-visually [-h] [--silent] [--verbose] [--threshold THRESHOLD]
                         [--dpi DPI] [--time TIME]
                         a.pdf b.pdf

Compare two PDFs visually. The exit code is 0 if they are the same, and 2 if
there are significant differences.

positional arguments:

optional arguments:
  -h, --help            show this help message and exit
  --silent, -q          silence output (can be used only once)
  --verbose, -v         show more information (can be used 2 times)
  --threshold THRESHOLD
                        PSNR threshold to consider a change significant,
                        higher is more sensitive (default: 100)
  --dpi DPI             resolution for the rasterised files (default: 50)
  --time TIME           number of seconds to wait before discarding temporary
                        files, or 0 to immediately discard

These “temporary files” include a PNG image of where any differences are, per page, as well as the log output of ImageMagick. If you want to get a feeling for thresholds, there are some example PDFs in the tests/ directory.

There is also an environment variable:

  • COMPARE: override the path of ImageMagick compare. By default, we try first compare and then magick compare (for Windows).

So what do you use this for?

Personally, I’ve used this a couple of times to refactor my LaTeX documents: I just simplify or remove some macro definitions, and if nothing changes, apparently it’s safe to make that change.


At the moment, this program/module works best for finding whether two PDFs are visually different.

This project will not work on Python 2.

The code is dual-licenced under both

at your option.

Supported Python versions

The versions that are regularly tested can be found here, that’s probably Python 3.8 and Python 3.9.

For your convenience we declare more Python versions acceptable in pyproject.toml, but the non-tested versions could potentially break from time to time. My goal is to support basically Python 3.x; please let me know if something doesn’t work on an older version.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

diff_pdf_visually-1.8.0.tar.gz (86.9 kB view hashes)

Uploaded source

Built Distribution

diff_pdf_visually-1.8.0-py2.py3-none-any.whl (15.1 kB view hashes)

Uploaded py2 py3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page