A webcast differs from traditional television on several important dimensions. This section describes some of these differences and develops a production model for webcasts. The design and implementation of dc reflects this production model. We begin by examining the types and number of data streams that may be involved in a webcast. Next, the source infrastructure used in a webcast is described. Finally, the receivers of a webcast are characterized. In each case, differences between the webcast environment and a conventional television broadcast environment are highlighted and used to motivate developing tools that meet these webcast-specific needs.
Unlike a television broadcast that is limited to one video and audio stream, a webcast can be comprised of multiple video and audio streams along with associated hypertext documents. The number of streams may also be dynamic. For example, during the webcast for the Berkeley MIG seminar, one video stream may be initially used when the speaker is presenting his material and another video stream showing the audience may be added to the webcast during the question and answer period at the end of the seminar. In a traditional television broadcast, switching between a view of the speaker and a view of the audience might be used during the question and answer period. This technique could also be used during a webcast, but providing both streams simultaneously is another possibility because webcasts may have more than one video stream.
The infrastructure available to a webcast is an IP-based data network with cameras and microphones attached to PC's that capture, digitize, possibly compress, and transmit the media streams. A television broadcast, in contrast, requires expensive, special-purpose hardware operated by trained technicians. The sophistication of a webcast production environment may vary from a single video and audio stream to a situation like the Berkeley MIG seminar where multiple video and audio streams, an effects server, remote participants, and auxiliary hypertext streams are used. The tools used to produce a webcast are required to accommodate heterogeneity in the production infrastructure.
The receivers of a webcast are also heterogeneous. For example, in the Berkeley MIG Seminar, a high quality version of the webcast is sent to viewers on the local campus and Internet2 connected sites that have high bandwidth connections and experience little delay. A low quality version of the webcast is sent to Public Internet viewers that may have limited bandwidth connections. The director must manage two different transmissions of the same material. These transmissions may differ in bandwidth and quality and may also differ in the number and type of streams sent. In a television broadcast, only one version of the program is transmitted and the capabilities of the receivers (i.e., television sets) are standardized.
Dc was developed to manage and control the production of live webcasts. It uses a general webcast model composed of three stages: source, webcast, and transmission as shown in figure 3. Sources are the streams available in the studio session. From this set of sources, a subset is selected for the broadcast. The speaker and content stream, for example, can be chosen from the video sources in the classroom or any source available in the studio session (e.g., speaker camera, audience camera, stored material, or presentation PC). Finally, multiple copies of the webcast, called transmissions, are produced using different technologies (e.g., Real Networks, Windows Media etc.) and transport parameters (e.g., bit-rates and formats). These transmissions are selected to match the expected capabilities of the audience.
