Multimedia
Blogs
Forum
Affordable HD
Whitepapers
Newsletters
Events
Advertisers
Blogcast
MillimeterMedia Asset Management at Microsoft (registered trademark) Studios
Aug 1, 1998 12:00 PM, Kevin Comerford, Mark Beauchamp, and Robert Culver
In February 1997, the Interactive Media Production (IMP) group at Microsoft (registered trademark) Corporation implemented Bulldog One.Six as its organizational media asset management system. This event followed nearly two years of exhausting discussion, investigation, due diligence, and system planning-the goal of which was to create a reliable data repository where digital media content could be safely stored, tracked, and made accessible for re-use.
Media production has been a business activity of Microsoft Corporation since the early 1990s. Since that time, the company has established several internal production facilities to create media content for its diverse needs, which include consumer software products, online services, television broadcasting, corporate presentations, training, and research.
The Interactive Media Production group, which originally sponsored the asset management project, was formed in 1996 from several earlier production units. This group's mission was to be the company's central service bureau for digital media production. In May 1998, IMP combined with the internal Tele-production facility to form Microsoft (registered trademark) Studios-the largest internal media production resource in Microsoft's corporate history. This is where the Media Asset Management project is presently headquartered in Redmond, Washington.
The Media Asset Management Problem Over the past several years, digital media production has become an increasingly important part of software product development. However, as digital media usage increases, the challenges associated with managing it effectively also become more complex. Traditional methods of managing digital media files, whether they are stored on networked file servers or on high-capacity removable storage devices (such as optical discs or Iomega Jaz cartridges), typically work well while a product is in development. Over time, however, these solutions actually inhibit subsequent access to archived elements.
This is partly because removable storage systems take the assets offline and make restoration and retrieval a cumbersome, time-consuming process. Likewise, online file server storage can be equally impractical. The sheer volume of new elements created each day by a large production house makes it easy to lose track of older assets as they are moved or deleted to make room for new projects. Regardless of the storage system though, long-term media asset storage is universally problematic.
There is no permanent, reliable way to affix identification and descriptive information (here defined as metadata) to an asset or group of assets while being assured that the data will be available to later users. In effect, these shortcomings cause media content to become "digital mulch." Within a short time of a project's completion, it gradually becomes buried under layers of newer content and finally settles to the bottom of a file server directory structure-or worse, to the back shelves of an archive room. Ultimately, as production staff turns over, these elements become unidentifiable and lose both their financial and intellectual value to an organization.
Early steps In an attempt to leverage re-use of existing media content, Microsoft's asset management project was originally formed in late 1995 when a small team of production support personnel was assembled to develop an Interactive Television (ITV) archive system. This group consisted of Mark Beauchamp, S enior Data Engineer, Kevin Comerford, Media Archivist, and Robert Culver, Database Systems specialist. Even as the initial requirements for the project were being developed, the media production environment at Microsoft (registered trademark) was expanding and evolving at an aggressive rate.
It quickly became apparent that a highly specialized asset management solution was completely inadequate to meet the organization's needs. By mid-1996, the media production groups were already producing hundreds of new elements each month that ranged from small vector graphics and bits of HTML code to extremely large raw MPEG video streams. This content was being produced for a variety of Microsoft consumer products and designed for delivery on CD-ROM, the World Wide Web, and television.
Cognizant of the perilous virtual compost heap phenomena that befalls unmanaged storage systems, the asset management team identified three primary requirements for their system. First, it should provide a digital repository, where media assets could be maintained online and still be archived reliably. The repository should also accept any existing category or format of digital media and be extensible to accept new formats as they are developed.
Second, the system should provide an application with a clean, attractive interface that is easy for production staff to learn and use. It should also provide full support for previewing any digital asset archived into the system and include the ability to group assets into multiple hierarchically organized collections based on local needs.
Third, the system should provide tools for managing media production workflow and be able to link to other systems-such as accounting and intellectual property tracking.
This needs-assessment helped the asset management team decide early to not write their own application because of the enormous time and staff resources needed to fulfill the project's goals. Efforts then turned toward investigating finished third-party solutions. At that time, the media asset management marketplace was only just emerging, and available products were either very specialized in favor of one class of media (e.g., video) or designed for small workgroups and very lightweight media production volumes.
None of the applications examined satisfied every requirement, so we decided to perform an onsite evaluation of the three most promising systems in order to let the media production staff decide which tool best met their needs. A representative sampling of designers, editors, producers, archivists, and technical staff was invited to attend training on each application, use it in a practical trial, and then rate it against a structured set of evaluation criteria.
When the results were tallied, it was obvious that the software from Toronto-based Bulldog was the winner. Users indicated that they were definitely attracted to the Bulldog user interface, which was cleanly designed and easy to learn. However, there was also a consensus that Bulldog addressed the broadest range of asset management issues of any application then on the market.
Features that users rated most highly included the system's organizational capabilities, multi-platform support, asset version control, associations, automatic history tracking, automatic thumbnail generation, and batch import capabilities. The asset management team also found the Bulldog system architecture compelling and appreciated The Bulldog Group's willingness to collaborate with them to meet their specific system requirements. Based on these factors, Bulldog One.Six was implemented in a pilot program in early 1997.
Bulldog One.Six in Action Once Bulldog One.Six was installed and tested, the asset management project became centered on the Microsoft (registered trademark) Studios Media Archive group, which had to develop new workflow systems and procedures to begin importing its digital media collections into the database. The Archive receives media assets not only fromMicrosoft Studios' productions, but also directly from many of Microsoft's consumer software development groups. Each product group maintains its own production workflow and schedules, so the Media Archive must tailor its services to fit the needs of each of its major customers.
Thus, there was fertile ground upon which to test the capabilities of the new asset management solution. During the first twelve months Bulldog One.Six was online, the Media Archive group imported and cataloged an average of 5,000 new media assets each month (or about 250 assets per work day), which included audio, video, still graphics, and animation elements. Asset file sizes ranged from 3MB bitmap images to 9GB video streams. More often than not, the Media Archive staff would run batch asset imports 24 hours a day, five to six days per week. Given these operational parameters, it was essential that the asset management system provide excellent performance as well as an efficacious set of processing tools to assist in funneling the deluge of new assets into the database.
The Bulldog Group worked closely with Kevin Comerford, the Media Archive Manager, to design a batch-registration interface that created a text import script of a group of files to be imported. Once the script ran, copies of the media files and their associated metadata were moved into the Bulldog database. As an extension to the batch registration process, Kevin was later able to build a Microsoft (registered trademark) Access applet, which generated import scripts using existing asset metadata from various corporate databases. This was a tremendous time-saver as it liberated the archive staff from much of the manual data entry required to catalog new assets.
Archival imports and data processing work has continued uninterrupted since Bulldog One.Six was brought online-a tribute to the robustness and stability of the product. However, the pace at which new material flowed into the Media Archive for disposition quickly began to strain the performance limits of the system. By late summer of 1997, it was clear that Bulldog One.Six' client/server architecture was not designed to support the number of users clamoring to use the system each day or perform multiple batch asset registration sessions at once.
Performance delays limited the processing rate for new assets and began to frustrate users. While this caused some measure of disappointment, there was understanding that Bulldog One.Six was a maturing tool with expected limitations on system performance and functionality. Knowing that Bulldog Two.Zero was in development, One.Six was accepted as an intermediate step toward the final objectives of the project.
The Upgrade Path: Looking Toward Bulldog Two.Zero In early 1997, the Microsoft asset management team went to Toronto to visit The Bulldog Group and assist in requirements gathering and feedback for Bulldog Two.Zero development. Peter DeVries, Bulldog's Chief Technology Officer, met with the team and discussed the enhanced system architecture, which focused on improving performance and scalability by distributing data and processes across the network. This would insure that the performance bottlenecks emanating from the Bulldog One.Six client/server architecture would be eliminated. The Two.Zero architecture also allowed for the separation of technical components by creating a unique database- and media-independent solution.
This provided an excellent opportunity for the Microsoft Studios asset management team to undertake a proof-of-concept port of Bulldog Two.Zero to Microsoft (registered trademark) SQL Server 6.5. Robert Culver, Database Systems Engineer, undertook this project in January 1998. By coordinating with the Bulldog Two.Zero Engineering Team-who provided access to the Bulldog DBIS (Database Interface Server) source code-Robert had a demonstrable prototype running on SQL Server by March. The results were very encouraging. The Bulldog Group brought the final development and QA for the SQL Server port in house and then released a Beta version of the product to Microsoft (registered trademark) Studios in May. A GA release is expected by the end of June 1998.
Bulldog Two.Zero also features a fully customizable DHTML (Dynamic HTML) user interface, which can be modified rapidly and inexpensively to match a customer's changing business needs. Several timesaving features are built into the new batch-registration utility. The interface makes it easy to drag-and-drop collections of files from a local system and organize the desired database structure prior to registration. During a batch import session, Bulldog Two.Zero automatically builds graphical key-frame storyboards for digital video streams and extracts detailed metadata for both audio and video elements-in addition to generating thumbnail representations for still images. Batch metadata updates can also be performed on existing asset records.
Media Plug-in (registered trademark) technology, an important architectural component of Two.Zero, insures the system can intelligently support new file formats and algorithms as they are developed. The key feature of Bulldog Two.Zero from a library/archival standpoint is the new graphical thesaurus tool, which lets each Bulldog installation manage a controlled keyword vocabulary online and provides a simple point-and-click system for adding keyword terms to asset records.
Final Notes The new features and robust architecture of Bulldog Two.Zero provide fertile ground for Microsoft Studios to expand its asset management services. The potential application of key Microsoft technologies-such as NT5.0, SQL Server 7.0, Microsoft Transaction Server, NetShow, and COM+- will continue to be of mutual interest for both companies. The diverse needs of a growing base of corporate and consulting clientele will undoubtedly encourage collaboration.
Winston Churchill, the original "Bulldog," said, "Heroes pursue success as the outgrowth of inner values." In a similar regard, The Bulldog Group has definitely been an asset management hero for Microsoft (registered trademark) Studios. Bulldog's inner values have been well represented by their commitment to building a quality product that will have continued relevance in the rapidly evolving digital media industry.
Kevin Comerford, Media Archive Manager, Mark Beauchamp, Chief Systems Engineer, and Robert Culver, Database Systems Engineer, work for Microsoft Media Services, Redmond, Washington.
Microsoft Studios is a full-service multimedia design and production facility. It provides complete audio, video, and graphical production services for Microsoft's corporate communications, retail products, advanced research, sales, and marketing activities. In addition, its staff provides technical consulting on emerging digital media technologies such as DVD, compression, and digital television.
Continue the discussion on “Crosstalk” the Millimeter Forum.
