Digital Amphitheatre
Building the User Interface
A key aspect of the digital amphitheatre is its innovative user interface. This is implemented based on the vic video conferencing application from Lawrence Berkeley National Laboratory working in conjunction with the UCL Robust-Audio Tool.
The user interface of vic has been augmented with an additional mode, which displays the speaker, four panelists, and a number of audience segments (in our present implmentation, six segments are used) in a full-screen configuration. Due to the use of spatial tiling agents, it is only necessary to display a small number of video streams for the audience: each stream contains a 5x3 block of 80x64 pixel frames. The result is an interface that displays hundreds of video clips, while only receiving a small number of RTP streams (our present implementation receivers 11 streams in total, yet displays 95 video sources). The figure below shows the combination of streams via the spatial tiling agents, and their composition in the user interface.
Due to the use of background substitution, the interface is clean and simple. It displays no information about each source by default: a tool tip popup is used to highlight participant names and other information (the mixer functionality included in RTP allows for this to be conveyed along with a tiled video stream).
In present implementation, those participants which appear in the panel and as the speaker are selected command-line parameter, referenced by RTP CNAME. This provides for static selection of panelist and speaker. We plan to add the ability to dynamically chose panelists in a future version.
Audio is provided by a separate instance of the UCL Robust-Audio Tool, running on a local host. All communication and coordination between audio and video - including lip-sync - is achieved via a message bus interface.