MediaCrush [1] is not a Flickr, YouTube, or SoundCloud stand-in, nor is it meant to be. If that's what you need, check out MediaGoblin, which I talked about in a previous article [2]. MediaCrush works more like Imgur [3], which allows you to upload images quickly and easily – with no registration required (Figure  1).

Figure 1: Upload images quickly and easily in Imgur.

Like Imgur, MediaCrush (Figure 2) does not require registration, and it is absurdly simple to upload images to the site. The main aim of the site is to provide links or code snippets for embedding your media elsewhere. It was originally developed specifically for Reddit and is 100% RES (Reddit Enhanced Suite) compatible [4].

Figure 2: MediaCrush is similar to Imgur but has a number of neat tricks up its sleeve.

Even better, MediaCrush is completely open source, distributed under a permissive MIT license. Apart from bitmap images, it allows for vector SVG graphics, video, and audio, and it uses all the tricks in the book to compress your files so your site loads faster for visitors.

For example, when you upload a GIF animation, MediaCrush converts it into a losslessly compressed video (OGV or WEBM) and then shows the video on your site. A 998KB GIF I tried was compressed down to a 295KB OGV video – a 338% improvement (Figure 3). This has the added advantage of allowing users to pause the animation at any point. If you download the file, MediaCrush serves the original GIF image.

Figure 3: MediaCrush turns a GIF into a compressed OGV, making it a fraction of its original size.

Get Ready

MediaCrush comes with a very comprehensive README.md file with instructions on how to install the software. Unfortunately, not everything works as described.

As my test machine, I used a stock Debian Wheezy on VirtualBox with all updates applied as of November 26, 2013. When installing Debian, I chose the server option (databases, web servers, etc.).

The first thing you want to do is to get rid of Apache, so it doesn't hog port 80.

/etc/init.d/apache2 stop
apt-get purge apache2

Next, make sure your repos are up to date with no obsolete software hanging around:

apt-get update
apt-get upgrade

Because you'll need it to get the MediaCrush source, you have to install Git. While you're at it, you might want to grab Vim, too, which is useful for editing all the config files later.

apt-get install vim git

The MediaCrush developers also recommend compiling and installing the latest version of ffmpeg to manage and convert video files. Before that works, you'll have to install more packages from your repositories:

apt-get install nasm libtheora-dev libvpx-dev ↩
   libx264-dev libvorbis-dev pkg-config

Now, you can install FFmpeg, as shown by the command lines in Listing 1.

Next, you'll want to install the MediaCrush-specific dependencies:

apt-get install redis-server jhead tidy optipng

and Python-specific dependencies:

apt-get install python-dev libpcre3-dev python-virtualenv

Your system is now ready to install MediaCrush.

Getting the Code

To store the source code you're going to get from GitHub, create a directory,

mkdir mediacrush

and then get the source:

git clone http://github.com/MediaCrush/MediaCrush </route/to/your/>mediacrush

The download shouldn't take long. Once it's finished, create a virtual environment for MediaCrush:

virtualenv </route/to/your/>mediacrush --no-site-packages

This command ensures that MediaCrush uses only the modules it needs and that it also does not interfere with the rest of the system. To activate the environment, go into your folder and enter:

$ cd </route/to/your/>mediacrush
$ source bin/activate

Now you have to install the Python modules that MediaCrush needs. The requirements.txt file contains a list of all modules Python requires. Using pip,

$ pip install -r requirements.txt

you can install everything in one go.

Get Going

Before running the server, you need to do several things. First, you have to create a storage directory that will store the uploaded files under your MediaCrush directory, such as:

$ mkdir storage

Second, you have to rename the config.ini.sample file to config.ini and configure it. One of the changes you'll have to make in this file is at the storage_folder line. For everything to work smoothly, you have to change

storage_folder = storage

to the absolute path to your storage directory:

storage_folder = </route/to/your/>mediacrush/storage

Additionally, you have to change the protocol and domain key-value pairs. Although the official MediaCrush site uses the HTTPS protocol to encrypt traffic to and from the site, for the sake simplicity, change the protocol value to http. For domain, you enter the domain part of your site; for example:

domain = mymediacrush.net

To suit your setup, you may make many other changes in config.ini. For example, you can integrate your Google AdSense and Analytics accounts, include the email of an administrator for abuse reports, or change the location of directories and files.

Finally, you have to run the MediaCrush daemon that does all the conversion in the back end:

$ python daemon.py&

The & at the end pushes the execution of the daemon to the background, allowing you to run the server up front in the same terminal. If you enter

$ python app.py

MediaCrush runs in debug mode on port 5000. If you point your browser to the IP address of your server (e.g., http://192.168.1.110:5000), you should see a spanking new MediaCrush install waiting for your pics.