German broadcaster ZDF has focused its new digital archive system (DAS) on the newsroom. There was a legacy issue to solve involving the 1.3 million documents generated since 1992.
Whether you call it content management (CM), asset management (AM), media asset management (MAM) or digital media management (D-MM), the issues confronting broadcasters in storing and retrieving program material have become global issues. The task has become more complex, making traditional program production technology insufficient to these new challenges. Fortunately, broadcasters have been able to obtain the new capabilities by piggybacking on commodity products from the IT world. From disk drives to networking systems, the IT community can provide a wealth of solutions to broadcast applications. And, the advantages are several: ready supply, low cost, multiple suppliers — and a relatively mature technology base.
The ideal solution is not yet clear, and, in fact, there is no single ideal solution. Couple this with the hesitancy of broadcasters to adopt anything not birthed from their own community. IT-based solutions have only recently found welcome in our closed community. Fortunately, there are many vendors now offering a range of working solutions based on IT technology.
Several papers at the IBC2003 Conference confirmed that progress of moving to IT-based solutions is being made, and research is continuing to refine the process. If you visit any broadcast newsroom, chances are the word you hear most often is “where.” “Where's my stuff?” “Where's the clip?” “Where's the script?” “Where's the running order?” In short, the answer to the age-old question “Where's my stuff?” is still at hand. Let's look at two proposed solutions.
Networking solutions
The first example was described by consultant Juergen Heitmann of AV Media Technology (Germany). He described content management activities by several broadcasters in Germany.
Heitmann noted that hardware and software elements are already common in most broadcast products. What broadcast system doesn't already use a hard drive, display and Windows software? But an IT-based production platform means more than just the use of such equipment. It means the use of regular IT hardware products in combination with broadcast specific software. In this case, the task of content management is supported by common IT-based technology networks transporting the content in the form of files. This allows almost any number of people access to the material all at the same time.
The thing to acknowledge before the adoption of a network-based broadcast system is that there will be a change in workflow. Disregarding the human issues with changing things, it also means that a thorough, logical analysis of all processes and cross-links must be carried out. At present, ZDF German Television is into the gradual installation of a digital archive system (DAS) for the program inventory stored in its production and distribution center in Mainz.
It is not a “typical” digital archive system. For one, the video, audio and metadata are delivered to the archive as a data file transported via IT networks. Building a network system was a top ZDF priority. Then there is the legacy issue. ZDF had to integrate the new DAM system into an existing TV program database that contains more than 1.3 million documents, plus legacy software control modules. Thus, the system had to offer an open solution with precisely defined interfaces to make the integration of legacy systems possible. Also, ZDF required a “distributed” solution that integrates systems from different locations. Finally, scalability of the architecture is key so that the system can grow along with programatic demands.
In another example, two broadcast operations, SAT1 and ProSieben/N42, needed to merge into ProSiebenSAT1 Media AG. This required the integration of the system environment used by ProSieben/N24 in Munich into the existing infrastructure of SAT1 in Berlin. Not only were two systems brought together, but also it was done without impairing the ongoing broadcast operations of either facility. The result is the modification of two divergent legacy systems into a single asset management system.
The learning curve has been steep. However, while Heitmann says that the fully IT-based television production facility is not yet here, he adds that it's “no longer a question of ‘if,’ but a question of ‘when and how.’”
Finding video with MPEG-7
In this digital era, there must be a better way to retrieve video than a text-based search. It is often a hit-or-miss proposition, depending on how the user describes the video to be found. Why not query the video being searched for using the video itself? While that sounds like a strange way to go about it, researchers at KBS Technical Research Institute in Korea have done just that using a system based on MPEG-7 technology.
In their IBC Conference paper, B. H. Jung, M. H. Ha and K. S. Kim point out that the traditional Query-By-Text method has several basic problems when applied to a multimedia information system.
The first is that a lot of manpower and time is required to produce a useful text description of each scene or clip. The system designer has to provide several description-appropriate labels, such as title, source, genre, character name and actor name, that are appropriate to the application. Then, at the ingest stage, someone must input the proper data to fill these fields. If the fields aren't properly completed, the asset may not be properly categorized or may even be lost. Another problem is that different users may use different words to describe the same clip.
Using a Query-By-Example (QBE) approach developed by the authors, users are able to quickly locate desired media without using text-based searches. A QBE retrieval is based on an algorithm that ranks the database clips according to a predefined measure of similarity between the query and the target. The KBS system consists of three main components: MPEG-7 metadata generator, query processor and query client.
The system uses a traditional IT client/server architecture. The client is a query terminal, and the server is the query processor. When the server and client send and receive the query and the result, they exchange information using common XML language (extensible mark-up language). (MPEG-7 expresses its metadata — MPEG-7 Ds or descriptors — in terms of XML.)
The metadata generator detects the scene/shot boundary and extracts features such as color and motion activity automatically from each shot. There is an MPEG-7 D visualization module to confirm the extraction result. Because MPEG-7 D itself is not understandable, it is difficult to validate the result. So the metadata generator not only generates the MPEG-7 descriptors, but also helps the user to validate the extracted metadata value. The generated metadata is stored in the XML document or exports to the database. The database scheme is made according to the MPEG-7 descriptor hierarchy. The query processor server's role involves the comparison of features between the video in the database and a query video. The query can focus on an object within the frame or the full screen. A user can specify the query's parameters by adjusting the weighting factor of each feature. Then the server responds with the shots most similar to the query.
KBS plans to refine the QBE system by adding a character retrieval module and a Web-based video retrieval system. The researchers suggested that advanced object retrieval and much faster retrieval speed are future projects.
Such a retrieval system would be extremely helpful in fast-paced news environments, where B-rolls are often needed to support a story and must be located quickly — often by the less experienced users. By needing to merely define certain video aspects of the clip, a content management system like this would provide tremendous benefits to such environments.
Is tape (finally) dead?
Since the dawn of videotape, archives have been a chief victim of the format wars. In the last 25 years, some 33 videotape formats have been used by broadcasters, not counting consumer formats and the data formats favored by the computer industry. (There was also the BBC's ill-fated VERA.) The videotape market also had another pitfall for archivists: What if a manufacturer stopped making VTRs? One prominent European broadcaster once put 2-inch quad machines in mothballs on the possibility of needing them for archive material.
The various drawbacks of videotape archives are well-known. Tape is intrinsically fragile and requires expensive environment storage conditions. Under ideal conditions, tape's life expectancy is only about 25 years. The record/replay technology is expensive. In addition, tape robotic systems are physically large and complex. And access times are relatively long, even with high-speed shuttling. Otherwise, tape archives are fine.
Ted Taylor, Panasonic Europe, presented the viewpoint that optical disks are the perfect solution to long-term storage. Taylor noted that like VTR formats, the number of optical disk variants currently and soon to be available is becoming almost bewildering. Fortunately, they all have a degree of physical commonality: All are 12cm in diameter, will probably reside in a universal caddy and are scanned by a laser with relatively simple, inexpensive mechanisms.
The big breakthrough for optical disks, of course, was the development of the DVD. And now Blu-Ray variants are appearing that employ a shorter wavelength (405nm) blue laser to improve both storage capacity (23GB/25GB/27GB) and data transfer rates (36Mb/s). The author claims that future developments are in the works that will make optical storage an even more attractive proposition:
- Multilayering techniques to increase capacity on Blu-Ray (2004).
- Blue MAMMOS with 100GB + 100Mb/s (2005) and 200GB + 200Mb/s (2007).
- Holographic disks with 200GB + 160Mb/s (2005) that do not need to spin during write/read.
Another obvious advantage of disks is a robust life expectancy of 80 to 100 years without the need of controlled environment conditions. In addition, reader/writers are inexpensive devices, and disk robotic systems are relatively simple and less expensive than tape cart systems. Finally, and not inconsequential, disks provide rapid and random access.
As for the financial case for disks, Taylor claimed it was no contest. Optical disk storage costs about one-third as much per program hour stored compared with tape-based options. (See Figure 1.) And the great financial advantage of optical disk storage is that it is a consumer-based technology. Therefore, costs can be expected to drop dramatically as manufacturing volumes increase.
