Thoughts on Future Requirements for Video on the Web

by Eric Hyche, RealNetworks

W3C Video on the Web Workshop
December 12-13 20007

When thinking about the future of video on the web, it is useful to take a bottom-up approach in considering the various components that go into web-delivered video:


Several years ago, it seemed that the issue of video quality was going to be the defining issue in the choice of codec for web video. Proprietary and standard solutions played a continual game of leapfrog, each attempting to set the standard for the best video quality at low bitrates. However, as broadband penetration increased, the bitrates used to encode web video have increased as well, moving out of the bitrate areas where large distinctions can be made between codecs and into the area where most latter-generation codecs look "good enough". Certainly the success of such User Generated Content sites such as YouTube have shown that for user-generated content, the quality of the video is not the most important factor. Important requirements in the future for codecs may be: a) better quality encoding on low-resource devices such as cell phones; and b) the ability to automatically generate searchable meta-data at encode time; c) the ability to deliver multiple bitrates from the same encoded stream, rather than have separately-encoded multiple bitrate streams in the file format.

File Formats

While video playback on devices has tended recently to gravitate towards standards-based file formats such as the MPEG-4 file format, these file formats often do not work as well for streaming progressive-download delivery. File formats that natively support delivery of multiple bitrates from streaming servers will continue to be important, as content providers often want to stream the same video file to multiple clients with diverse bandwidth connections.


It has become clear over the last few years that the requirements for a desktop video playback application and the requirements for web video playback are much different. Web video players must have seamless install experiences - users simply do not want to have to make a separate step to install a desktop player before they can watch a video on the web. However, desktop players can assist in offline playback by providing facilities to download web video.


The recent UGC video phenonmenon has shown that it's often difficult to both allow users to submit video to be shared and to respect ownership rights at the same time. More work needs to be done in finding ways to allow users to add their own value-add to content that they do not own. I often think of two use cases that I have been personally involved with: a) wanting to add a secondary commentary track to movie trailers for movies that I look forward to seeing; and b) wanting to annotate or highlight certain sections of video from clips of video for which I do not own the rights. Integration technologies such as SMIL should be useful for these tasks.