Video team sprint notes
Contents
An assessment of the current situation
- Current setup
- talk capture
- 2 cameras
- slide/presenter's laptop capture (goes to projector)
- hdmi (audio and video)
- vga (video)
- audio
- DP/miniDP to hdmi (audio and video)
- up to 2 speaker mics
- up to 2 audience mics
- up to 2 ambient mics
- three outputs
- talk capture output to streaming
- projector screen (shows slides/laptop)
- confidence screen (duplicate of projector for presenter)
- two inputs
- mixed capture
- remote source (e.g. streaming from another room)
- talk capture
We use TwinPacts and laptops running DVSwitch. These are getting old, don't support HD, don't support the modern HDMI laptops that presenters turn up with.
Sources of inspiration
Other places that capture presentations
- CarlFX
- Chaos Communication Congress
- Linux.com.au
?PyCon
- Quadrant
- FOSDEM
Thursday session
Objective 1: replace twinpact
Goals
- replicate existing functionality
- increase resolution to 720p or 1080p modes
- independent confidence screen for presenter
- remote video source (for repeater rooms)
- audio in/out tied to source/sink
- direct to ethernet or sdi
Discussion points
The consensus seems to be to use HDMI2USB in some form.
TwinPact replacement design proposal based on HDMI2USB firmware loaded onto a Numato Opsis connected to a Beaglebone Black. Three Opsis boards are on order for the video team.
We are currently prototyping on a Digilent Atlys
Objective 2: replace DVswitch
Goals
- replicate existing functionality
- increase resolution to 720p or 1080p modes
- select output on twinpact replacement
- select source for output #2
- record all inputs
- record all edits
Objective 3: streaming
Goals
- single domain name, website to start debconf video
- point "local" capture to streaming services
- log stats
- list and name streams
Discussion points
avoiding pulling by infrastructure outside of the conference venue to reduce problems with venue nat and firewall. Streaming architecture proposal
Objective 4: AV kit
Goals
- rearrange kit by room and package up on per-room basis
- cameras
- hdmi/sdi/ethernet
- lens
- mics
- licences for radio frequencies to run wireless mics
- DI boxes
- director talkback to camera operators
Discussion points
- The mixing station could be in one flight case that you open, connect up, and use.
- how much redundancy do we think we need?
- PCs are easy to obtain, we could keep spares of the hard-to-source capture cards
- video transmission protocol between camera and mixer
Decisions
- Kit by room
- Kit for a talk room
- 2 speaker mics
- main camera
- audience camera
- optional fixed wideangle webcam
- capture workstation with single set of capturing cards
- 2 ambient microphones
- sound mixer 4-channel
- Kit for plenary
- as talk room
- 2 more speaker mics
- 2 more cameras
- second set of capturing cards in workstation
- Kit for a talk room
Objective 5: kit for BOF room
Goals
- to be investigated
- consider videoconferencing kit as alternative to camera and mic
Objective 6: logistics
Goals
- can we avoid shipping cables?
- hire versus buy
Friday session
Objective 1: replace twinpact
Achievements
Lots of work on the Atlys-based prototype and software to support it. This will be tested at the talks of this Minidebconf.
Objective 2: Replace DVswitch
Achievements
People are looking at gst-switch and voctomix.
Actions to take away
- David Leggett to investigate Polycom videoconferencing equipment for objective 5
- Andy Simpkins and Andy Cater to create a software to generate a sponsor loop from arbitrary still images and package and upload it to the Debian package archive.