Videoplatform

Participating students: Andre Freitag, Andreas Hubel.

When I was a student I had the opportunity that once one of my advisors could not make it and my talk was recorded. Watching the video later was a very impressive and direct feedback channel. Seeing what people tell one would do makes a strong difference to just hearing and possibly not believing it.

Therefore I wanted to share this opportunity with students at the Technische Universität München.

In 2008 I started the project to create a system for recording and multiplexing a screencast of the slides with an HD record of the speaker. The backend is operating since 2009.

Since 2012 we also have a very nice frontend that is directly integrated with the DFN authentication provider for user authentication.

We regularly visit student's presentations and process them semi-automatically to finally publish them on the portal.

The system operates in multiple stages:

• The system is automatically parsing our announcement feed to create folders and meta data on the backend server.
• After recording with a screencast software and an HD camcorder the files are copied in the folders on the backend server.
• The backend produces low quality videos for faster manual cutting and synchronizing.
• After the synchronization the system automatically processes the HD videos according to the cutting information on the low quality videos.
• The talk is automatically published to the frontend together with meta data like abstract, slides PDF, or documents belonging to the talk.
• The speaker automatically gets an email with tokens that allow him to choose the visibility of his video which is only visible for himself by default.
• On the frontend it is possible to search for videos, rate videos, ask for permission to watch restricted videos, watch videos one can see without or after login, etc.

Have a look at the portal at http://media.net.in.tum.de/.
You can find a publicly visit talk series here.

More documentation and the scripts will follow soon (planned: end 2013).