Skip to content

Latest commit

 

History

History
37 lines (23 loc) · 2.56 KB

README.md

File metadata and controls

37 lines (23 loc) · 2.56 KB
Raw

imagesqueeze Travis CI build status

An image-optimisation app, currently used for comparing Guetzli with other compression standards.

Installation

imagesqueeze requires the Guetzli binary to be installed on your system. On OS X the easiest way to do this is through Homebrew.

  • Install Homebrew
    • /usr/bin/ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"
  • Once Homebrew is installed, use it to install the Guetzli binary.
    • brew install guetzli

Once Guetzli is installed imagesqueeze runs on node.js. If you've got Node and NPM installed than a simple npm install will download all of the required modules.

Basic usage

imagesqueeze's folder structure is pretty basic, containing an input and output directory (ignore www for now, that's coming later). Place any images you want optimized in input. imagesqueeze will run on anything in that directory, grinding it through the selected processes, and place final optimised images in the output folder, maintaining the original folder structure*.

*guetzli still has some issues with this so you may get an input folder in your output folder. Think of it as a delicious Reese Peanut Butter Cup.

To run Guetzli on the input directory:

  • grunt g

To run standard web optimisation (optipng, gifsicle, etc.) on the input directory:

  • grunt i

Known issues

  • Guetzli only works on JPEG formatted images. It has a nasty habit of also running on .png files but changing the extension which caused some errors, so grunt g will currently only pick up images with a .jpg extension and ignore the rest. grunt i will pick up everything.
  • Guetzli can only work on images in a certain colorspace. If it comes across a JPEG it can't handle for whatever reason the process stops. I haven't found a way around this yet.
  • Guetzli is sllllllllllloooooooow and will gorge itself on your computer's available memory like Oprah Winfrey at a buffet. A folder of images that grunt i can churn through in about 10 seconds can take grunt g several minutes. If you've got a bunch of images to work through maybe set it to run and then go have lunch.

TODO

  • The Gruntfile has a bunch of other in-progress bits in it (exporting output as a .zip archive, setting up a web app to run this online, etc.) that are all in various stages of completion but nothing's ready for prime time yet. Soon! Maybe!