The video team live-streams DebConf sessions over the Internet to remote participants. The sessions are also recorded, transcoded, and published in the Meetings Archive.
The team relies on conference attendees to volunteer to operate equipment. Video is live edited during the conference sessions, for both streaming and recordings. This allows the recordings to be published during the conference, without much manual post-production.
The three main rooms (2 talk rooms and a BoF room) had video coverage, and almost every event held in them was streamed and recorded. Due to technical problems, 2 BoF sessions weren't covered, and a few other talks had recording gaps.
 Preparation and Equipment
Debian only owns enough audio and equipment to cover 2 rooms, and even then, only just. Equipment has failed over the years, and hasn't been replaced.
The video team uses a DV workflow, with capture over Firewire. Unfortunately, this is an obsolete technology. Compatible cameras and laptops are becoming hard to find.
To fully cover 3 rooms, equipment had to be borrowed, bought, and hired. A friendly university provided 5 cameras, 2 audio mixing desks, and some microphones. FOSDEM provided frame-grabbers. Collabora provided a bag of retired laptops, that would be donated to a good cause after the confence. And some attendees provided their spare personal laptops.
The video team ran a separate private network to stream digital video (DV) from cameras and frame-grabbers to director's mixing desks (running dvswitch). From there to disk and stream encoders, and eventually out to the DebConf stream relay network on the Internet.
On Monday of DebCamp, the networking team had a network up, and provisioned a private network for the video team, from each talk room to the server room.
Laptops were provisioned, using the video team's standard FAI automated setup. During the next few days, all the talk rooms were set up, and streams brought up.
Each room received two cameras. One for tracking the speaker and another for anyone asking questions from the audience.
For sound, each room was equipped with two wireless headset microphones for the main presenters and 2 wireless hand-held microphone for other presenters, introductions, and audience questions. Except for the BoF room where we used a shotgun microphone on each camera. This required cameras to be pointed at speakers, rather than passing microphones around.
A pair of wired “ambiance” microphones gave the video a less sterile sound. They also caught some questions and audience discussion from people who did not wait for a hand-held microphone to be brought to them.
Two streams were provided: A high-bandwidth (1.5Mbps) SD h.264 stream, and a low-bandwidth (200kbps) QVGA theora stream.
Due to the small size of the main talk room, plenary events were live-streamed to the second talk room, to provide overflow seating. There was provision for a stream back to the main room, for questions from the second room, but it was never needed.
 Southampton Solent University, England
The video team was dormant until quite late before DebConf, leading to a mad scramble to find equipment in time. That said, we succeeded, and got set up at the venue quite quickly.
A large number of volunteers are required to operate equipment during the conference. 58 Volunteers volunteered for almost 600 tasks. And in the evenings, another team of volunteers reviewed the recorded video of the day, for encoding and publishing.
During the conference, we recorded 1.6 TiB of DV footage. After cutting and editing 120 videos were produced and published in the archive adding up to over 70 hours of video.
 Future of the Video Team
Most of the technical problems faced by the video team during the conference were to do with unreliability of firewire ports and cables. Firewire and DV are really starting to show their age, and the team needs to find new workflows.
A sprint by the video team will be held later in 2015, to plan for the future, evaluate suitable software systems, and select hardware for purchase.