Crunch is a tool for lossy PNG image file optimization. It combines selective bit depth, color type, and color palette reduction with zopfli DEFLATE compression algorithm encoding using the pngquant and zopflipng PNG optimization tools. This approach leads to a significant file size gain relative to lossless approaches at the expense of a relatively modest decrease in image quality (see example images below).
Historical benchmarks with the files included in Cédric Louvrier's PNG Test Corpus versus other commonly used PNG optimization software are available in BENCHMARKS.md.
Crunch PNG image optimization is available through the following applications in this repository:
crunch- a *nix command line executable that can be used on macOS, Linux, and Windows POSIX application deployment environments such as Cygwin- Crunch GUI - a native macOS drag and drop GUI tool
- Crunch Image(s) service - a macOS right-click menu service for PNG images selected in the Finder
Install and usage documentation links for each of the Crunch applications are available below.
The crunch command line executable can be installed with make or manually. Please see the Install documentation.
Enter paths to your PNG image files as arguments to the crunch executable. Please see the Usage documentation.
The Crunch native macOS GUI application can be installed with Homebrew or the dmg installer that can be downloaded from the repository releases. Please see the Install documentation.
Drag and drop one or more images on the application window to process your PNG files. Please see the Usage documentation.
The macOS right-click menu service "Crunch Image(s)" can be installed with make or manually by drag and drop in the macOS Finder. Please see the Install documentation.
Select one or more PNG images in the Finder, right-click, and select the Services > Crunch Image(s) menu item to process your files. Please see the Usage documentation.
The following examples demonstrate the benefits and disadvantages of the current iteration of Crunch's aggressive space saving optimization strategy. In many cases, the PNG optimization minimizes file size with an imperceptible decrease in image quality. In some cases, degradation of image quality is visible. View the horizon line in the prairie photo below for a demonstration of an undesirable artifact that is introduced with image processing. Experiment with the image types that you use and please submit a report with examples of any images where the image quality falls short of expectations for production-ready files.
- Original Size: 583,398 bytes
- Optimized Size: 196,085 bytes
- DSSIM similarity score: 0.001471
- Percent original size: 33.61%
- Original Size: 138,272
- Optimized Size: 66,593
- DSSIM similarity score: 0.000948
- Percent original size: 48.16%
- Original Size: 196,794 bytes
- Optimized Size: 77,965 bytes
- DSSIM similarity score: 0.002988
- Percent original size: 39.62%
- Original Size: 197,193 bytes
- Optimized Size: 67,596 bytes
- DSSIM similarity score: 0.000162
- Percent original size: 34.28%
- Original Size: 249,251 bytes
- Optimized Size: 67,135 bytes
- DSSIM similarity score: 0.002491
- Percent original size: 26.93%
- Original Size: 440,126 bytes
- Optimized Size: 196,962 bytes
- DSSIM similarity score: 0.000480
- Percent original size: 44.75%
All images above were obtained from Pixabay and are dedicated to the public domain under the CC0 Public Domain Dedication.
Have you identified a problem? Please create a new issue report on the Github issue tracker so that we can address it.
Crunch is licensed under the MIT license.
pngquant is licensed under the Gnu General Public License, version 3. The pngquant source code is available here.
zopflipng is licensed under the Apache License, Version 2.0. The upstream zopflipng source code is available here. The source for the modified zopflipng fork that is used in this project is available here.
See the LICENSE.md document for details and additional licensing information for this project.
Contributions to the project are warmly welcomed. Please suggest enhancements as new issue reports on this repository. Source contributors should fork the git repository and submit changes as a Github pull request.
Crunch is a simple tool that makes excellent, free, open source software built by others more accessible. The hard work on the optimization engines that run Crunch has been performed by:
- Lode Vandevenne, Jyrki Alakuijala, and the zopfli project contributors
- Kornel Lesiński and the pngquant project contributors
The fantastic macOS GUI animations were designed by Gary Jacobs.