OBSERVATIONS

The project was oriented on defining a set of end-to-end content-on-demand system requirements, developing a subset of well-defined components, and validating them with a focused enhanced application. Realistically, validating the entire system was far beyond the resources available to the Sony/WGBH team. Instead the team decided to implement the set of content-on-demand components sufficient to support an enhanced interactive version of the Antiques Roadshow application.

As with any development activity, there were many lessons learned throughout the course of the project. The observations of note for this project are stated below:

There are still many challenges to commercializing content-on-demand systems

-   Many systemic problems to overcome - carriage agreements, governmental regulations, IP infighting.

-   No commercially ubiquitous viable platforms are yet available.

-   Open requirements and standards must be developed for components and interfaces.

-   No standard way to generate and distribute metadata.

-   Content-On-Demand systems are more than simply VOD systems.

It is possible to build an exciting enhanced application

-   Providing a television experience with useful content manipulation was well received by viewers at every system demonstration including Broadband Plus.

-   The Antiques Roadshow producers were excited about the ARS demonstrator and want to study a test deployment.

Developing compelling enhanced interactive applications for television is technically attainable. Key features that provide the ability for exciting enhanced applications are

-   Content searching

-   Viewer preferences

-   Viewer controlled playlists

-   Links to extra content

-   VCR trick modes

The Metadata/DB is the key to providing effective, exciting user experiences

-   The database used by WGBH was very similar to the Sony DB which suggests a high degree of metadata commonality.

-   The user experience depends on quality and consistency of the metadata.

-   Metadata is the key to linking the original video program to "extra" information.

-   Though most metadata and "extra" information were being created for other sources, gathering/editing for enhanced use does add cost.

-   Thumbnail images used for metadata required additional per episode production tasks.

-   It does not require large amounts of metadata to develop compelling applications.

-   Automated metadata creation tools are needed to reduce cost and increase consistency.

The scope of the Antiques Roadshow application

-   Validated the functionality of an Antiques Roadshow application with one episode of clips.

-   Verified that the system can be built with web based technology.

-   Pointed out that scalability issues with metadata and content are important to resolve.

Platform portability is achievable through proper architecture and implementation

-   Some platform differences will exist but for the most part reusable components are possible.

-   The ARS demonstrator was ported to several one and two box client/server STBs. All ports were completed within a few days.

The UI is the drawing-card for the application

-   The UI must be tailored for a TV vs. a computer. There are many key differences between the two.

-   The look and feel must not conflict with the television video.

-   Ultimately, the goal is a passive "TV" like presentation of the content, where the viewer only interacts when desired.

PVR adds significant system capability and underscores manipulation features

-   Indexing into an episode and then playing for only a specified duration is an essential feature.

-   Synchronizing the Metadata timecode to actual recording time must be performed.

-   Current editing techniques (i.e. L-Cuts) produce awkward transitions when reordering content and current solutions increase the system cost/complexity.

-   Content and metadata can be distributed through multiple transports (i.e. broadcast/broadband); interactive version is only available once both arrive at the STB.

-   PVR allows alternate methods of viewing:

-   Explicit skipping to index/chapter points

-   Preferred

-   Manual playlist

-   PVR helps reduce the "distraction" of enhanced content to the dramatic arc of the original video program.

Flexible content searching is important

-   As PVR's have larger storage, searching recorded content is important.

-   Search should be able to search down to the segment level.

-   Search is only as good as the underlying metadata.

-   If the transcript (i.e. Close Caption data) is available, some metadata could be generated on the fly.

Preferences don't need to be complex to produce dramatic results

-   Simple preferences can create a valuable filtering/ordering mechanism.

Overall, the project goals were achieved. The next step would be to extend this work as a service proposition. This service opportunity should result in a limited field deployment of a service based on the Antiques Roadshow application on a deployable platform.

Revised Tuesday, 15-Apr-2003 07:08:32 CDT - h © 2000 - 2003 Local Enhancement Collaborative & .