What would make the Web a true video platform
Paper for the W3C Video on the Web Workshop, 12-13 December 2007, San Jose, California
- Author
- Loren Larsen, Move Networks
- Jeremy Condon, Move Networks
- Dave Brueck, Move Networks
Move Networks, Inc. is a leading provider of video services for premier content publishers. Move uses a proprietary adaptive streaming protocol utilizing standard web technologies to provide highly scalable streaming of live and on-demand HD-quality video.
Paper
Consumption of video is an increasing component of how users are spending time on the web. This was initially fueled by consumption of short-form video like YouTube, movie trailers, and news highlights. Advances in streaming technology are now fueling rapid increases in consumption of high-quality premium content of television and film assets. Users will consume all types of content and will demand their choice of where and when content can be consumed. This includes the ability to consume content on their computer, in the living room, and while mobile as well as consuming available content at any time of day or night. Publishers need to enable ubiquitous, yet monetizable access to content or users will find other avenues for consuming that same content.
So users are already watching video on the web at rapid growth rates, is there a problem here?
We propose to answer that question by examining the 4 basic steps in delivering video from the camera lens to the eyeball
:
- Encoding - Publishers will encode the asset into one or more video formats and store the video along with meta-data in one or more content management systems.
- Publishing - Publishing an asset involves not just making a single asset available with its meta-data, but assets may be organized into hierarchies (e.g. episodes of a show) or grouped into channels (e.g. a channel of "New York Yankees" world series triumphs). This is the step in organizing how one or more assets can be presented.
- Presentation - Web pages present the available assets in a variety of formats, but few standards have been implemented that enable page developers to represent available content in a standard way. Depending on the service, a page may need to pull content from multiple CMSs to access content from multiple content publishers. The video assets may not be in the same video format and may not be playable by the same media player. In addition player APIs are proprietary so the implementation varies depending on which media player is used to play the content.
- Distribution and Playout - This refers to distribution of available (meta-data) to the page as well as the content itself. The primary challenge we see in distribution is that of scale. Web-video in all its forms of consumption will be supporting 10's of millions of simultaneous viewers. The efficient distribution of these assets is critical to consistently delivering the end-user a high-quality experience. We know that quality matters and is directly related to the ability to retain viewers and monetize the captive attention of users. In addition, restricting access to content is an important part of monetization and honoring contractual obligations around premium content.
We believe consumers will gravitate to the most convenient and high-quality method for consuming the content. The web is the most ubiquitous mechanism for presenting and displaying video experiences to end users. In addition to some of the issues highlighted above we believe the following factors need to be considered :
- The web enables complex and rich viewing experiences where the video is one portion of a complete experience. Users may engage in community activities alongside the viewing experience and publishers may want to provide a level of interactivity ranging from text/audio commentary, live user commenting, closed captioned text or subtitles, video hotspots for interactive advertising, live game scores and statistics, synchronized web banner/video advertising, etc. The ability to synchronize each of these rich elements with video playback is critical.
- Web video today has primarily been about enabling viewing of pre-produced content. Live sports, entertainment and news events are increasingly becoming available and because of their timely nature these will drive special requirements. The first and most obvious of these is the number of simultaneous viewers. The second is live control of video playback, e.g. when to go to an ad break during a live event.
Consumers want to customize their experience on the web. This means finding and exploring the content that is relative to them. Video content needs to be easily searchable. The attributes that go with an encoded asset should be linked to the asset itself. Meta-data should allow search engines to locate thumbnail images representing video assets. Video should be linkable from web pages and treated as a first class object like images. For example the ability to link to a video asset at a particular time location might look like: <vid src="video.vid">. Unlike images, video has a temporal component and it would be helpful to reference from web pages certain ranges of video, e.g. <vid src="video.vid#30.5"> would play from 30.5 seconds into the video. Such a mechanism would enable anyone to easily create links to any video asset regardless of medium type or player needed to play it. If the ".vid" file also contained meta-data about the asset in a standard format, this would enable users to find the video they want and publishers to attract interested viewers. We believe it is valuable to include mechanisms for rating content so that users do not find content that has been deemed in appropriate (e.g. parental controls) and search engines are able to accurately identify traffic by its content.
We see the major work in making the web a video platform to revolve around:
- Creating a common framework for exchanging meta-data about video.
- Making video a first-class object in web pages
- Enable rights management by enabling users to authenticate themselves or their location.
- Identify scalable mechanisms for distributing meta-data with real-time updates for advertising, scores, etc.
- Identify scalable mechanisms for supporting 10s or even 100s of millions of users while ensuring high-quality video.
Until the issues above are solved, video is still a second-class citizen, and the full potential of using video on the web won't be realized.