AAF Gets Traction
Posted Aug 8, 2003

NAB 2003 marked a coming out party of sorts for the Advanced Authoring Format, a metadata interchange solution introduced in 1998. But despite mounting vendor support, widespread implementation has been slow to develop.

August 2003|It's been more than five years since Avid and Microsoft publicly launched efforts to define and promote the Advanced Authoring Format (AAF). AAF is intended to facilitate interoperability in the content creation workflow through standardized exchange of digital media and metadata. For the end-users of production tools in fields such as video post, broadcast, and audio post, the concept is a welcome one. But for vendors of the tools, implementation of interoperability is a complex issue involving both technical and marketing considerations. So progress toward standardizing and then implementing AAF hasn't exactly been swift. At this year's NAB, however, there were some hopeful signs that AAF might eventually become an industry-wide phenomenon, which makes this an opportune time for an update on the standard's status and outlook.

AAF's origins are in Avid's Open Media Framework Interface (OMFI), but it utilizes a "structured storage" container format developed by Microsoft. Implemented in an object-oriented C++ environment, it incorporates a vendor-neutral plug-in architecture, supports a broader range of media formats and temporal data, and allows the integration of 2D, 3D, text, HTML, and XML objects.

Beyond simply passing along media data (video, audio, graphics, animation, and text), AAF's emphasis is on interchange of metadata (data about the data), including compositional information that describes how sections of media data are combined and modified. It also supports version control, allowing an AAF file's data to be edited and revised while retaining the history of the changes, as well as media derivation information about the original source. Unlike formats such as QuickTime, AAF is not intended to serve as a delivery format, but rather is exclusively designed to serve the needs of the production community.

The increasing attention vendors are paying to AAF reflects the reality that today's production environments are diverse rather than monolithic, involving many specialized tools from multiple vendors. "The post-production market," says Rick Keilty, director of 844/X product marketing at Media 100, "includes a wide array of products and tools—video, audio, input/output, graphics, and 3D workstations—from a wide range of suppliers across multiple platforms, including Windows, Macintosh, Linux, and Unix, and supporting multiple formats such as DV, MPEG, and uncompressed 601."

This diversity is great when it comes to choosing the right tool for the job, but it creates a big challenge in terms of workflow. "Particularly with large projects," says Mike Nann, technical marketing manager for post-production at Leitch Technology, "collaborative and multi-platform workflows are the norm. Collaborative might mean between multiple users, such as visual effects artists working on different segments of a project, or between multiple types of users with distinct skills and specialties, such as editors exchanging project sequences with compositors and animators. It could be between multiple facilities, or even just a single user that might use two different compositing software packages."

What makes an interchange standard for production so appealing is the promise that this collaboration can work smoothly and efficiently, thus boosting productivity. "Collaboration and interchange are very critical in this segment," Keilty says, "in order to improve efficiency and thus provide cost and time savings. The entire content creation food chain has been pushed to deliver higher-quality, more visually sophisticated material in a shorter time frame for less money. Running around with tapes and paper copies of notes on editorial decisions is time-consuming, inefficient, and ultimately impacts design quality. So the need to share data between disparate systems and applications has reached a critical point."

The business pressures are occurring simultaneously with the long-term technological transition from linear to non-linear storage and production techniques. "The content creation industry," says Tim Claman, director of interoperability and standards at Avid Technology, "is moving away from workflows based on interchange of baseband signals and tape storage to workflows based on IP networks and storage."

As the technological foundation of the industry has changed, the methods available for facilitating heterogeneous workflow haven't kept up. "The complexity of today's post-production and broadcast workflows," says Paul Saccone, Final Cut Pro product manager at Apple, "makes it nearly impossible for any one vendor's products to handle everything. Seamless collaboration technologies are becoming a critical necessity. For years we've dealt with either inadequate or proprietary solutions."

The Collaboration Problem
"Fundamentally," says Christian Schormann, VP of engineering and chief strategist at Pinnacle Systems, "there are a few pieces to the problem of collaboration in the environment of IT-based video production systems. There is the interchange of media and the interchange of compositional metadata, meaning the information that explains how pieces of media are combined into a single production. And there is network workflow and asset management."

How do existing approaches to interchange measure up in each of these areas? "SDI has the widest compatibility," Schormann says, "but it is a pure push model of content; the receiver must take what is on the wire in a synchronous fashion. And SDI defines and transports very little metadata. In particular, there is no notion of compositions or compositional metadata. So the interchange is happening on such a low level that it is not particularly supportive for a content production workflow."

Edit decision lists (EDLs), on the other hand, do provide a form of compositional metadata, but Schormann says they are limited by their roots in the world of linear editing. "EDLs by far lack the semantic richness required to express the capabilities of modern editing systems. Also, they are not designed to operate properly in a world of networked digital storage where all media resides more likely in files on disk than on tape."

Avid's Claman agrees that an EDL-based approach is no longer sufficient. "The authoring tools have long since outstripped the capabilities of legacy interchange formats like EDL," he says. "AAF can express much more sophisticated structures than are available in EDL, AES-31, as well as OMF. I also see AAF gaining wider adoption than OMF."

As for the Pro-MPEG Forum's Media Exchange Format (MXF), Schormann says that AAF and MXF, being based on very similar container technology, are complementary. But while MXF "allows the inclusion of a healthy dose of metadata with the content, it is clearly not designed to be a composition format. It focuses heavily on providing a standard container format to interchange media. Also, compositional information, while it may be transported together with media at times, often needs a life on its own, divided from any concrete stream of media. MXF provides an industry-accepted standard format for media, while AAF carries the compositional information that goes beyond the reach of any given piece of media."

The emphasis on compositional metadata is crucial because, as described by Maurice Patel of Discreet, "post-production is all about modifying the source media to create a final product. These modifications, defined as metadata, are the added value of post-production. When many different tools are being used, the more data that can be exchanged, the more efficient the post-production process becomes. At the moment, metadata tends to be vendor-specific and cannot be transferred easily between applications. AAF allows a core set of metadata to be transferred quickly and seamlessly between different vendor applications without the user having to manually recreate the data."

Patel also likes the fact that "AAF is very powerful in that it is very flexible and can be customized to meet the specific needs of a wide range of different vendors. It is not a fixed or static format, but can adapt to changes in technology and production techniques."

Media 100's Keilty agrees that AAF has greater potential to adapt than other available approaches. "As toolsets become more sophisticated," he says, "the metadata required to allow for exchanging information between them has to also evolve. Thus, the focus on AAF, which includes a wider range of metadata definitions, as well as the ability to use ‘private' metadata, which will foster collaborative efforts between suppliers while preserving a degree of competitive advantage for a manufacturer."

Lastly, but not insignificantly, AAF's appeal is enhanced because it is seen as a true open standard. "AAF is not owned by any one company," Saccone says. "It represents the first serious effort of the broadcast industry to work together on collaborative technologies."

Varying Implementation
While acknowledgement of AAF's benefits seems near-universal, there's quite a range in the extent to which individual manufacturers have actually implemented the standard into shipping products. In fact, the question of implementation is a multi-faceted one because of the varying approaches and levels that might come under the umbrella of AAF compliance. "We can see various levels of AAF support today," Media 100's Keilty says, "all the way from just announcing future support, to only supporting AAF within a given manufacturer's product set, to needing to use third-party tools and translators to act as an AAF intermediary."

Keilty speaks of Media 100's approach— which involves AAF implementation in an unspecified future version of 844/X—in terms of prioritization. "Essentially, our priority would be development of an AAF-facilitated workflow in support of 844/X as a finishing system. The first step would be support for AAF compatibility between our own products. The second would be compatibility with the most commonly-used products from other manufacturers, and third would be use of custom-defined metadata within our product set or with other potential partners we may work with."

Avid's Claman, meanwhile, says that Avid regards AAF as a "core aspect of Avid's technology strategy. Most Avid products currently support AAF and we continue to actively implement AAF in our products." He adds that an AAF file exported by an Avid editor "can include" a variety of useful metadata, including sequence and clip information; edit decisions (how to reassemble a program from original sources); visual image metadata, including effects, key frames, and parameters values; user-defined descriptive metadata (e.g. "custom bin columns"); descriptive metadata for archival, and asset management applications; supers, titles, and subtitles; links to interactive content; and embedded audio media for sweetening.

Avid's AAF implementations are based on the AAF Toolkit SDK (free download at www.sourceforge.net/projects/aaf). "Avid products are not AAF-native," Claman says, "meaning they do not use AAF as the internal data model for storing data. For example, Media Composer still uses Avid project folders and bins, and Pro Tools still uses Pro Tools session files. There have been no real performance issues with implementing AAF because we use it as it was intended: as an intermediary interchange format."

Listing shipping Avid products that already support "full AAF import and export," Claman cites Media Composer, Xpress, Xpress DV, Symphony, Film Composer, MediaStation XL, NewsCutter XP, and NewsCutter Effects. Additionally, AAF import is supported in Avid DS and Avid DS|HD. And upcoming AAF capabilities for Digidesign Pro Tools and Softimage XSI were previewed at NAB.

Also upcoming (scheduled to ship by the time this article is published) is an optional AAF/OMF Interoperability Module from Leitch that will add import/export capabilities to the company's dpsVelocity 8.2 and dpsVelocityQ 8.2 nonlinear editors. "This is just the first stage of our plans to incorporate AAF," says Mike Nann, "and supports basic metadata such as timeline structure, tape descriptors, file descriptors, speed changes, volume adjustments, and SMPTE transitions. Our priority was to first meet the basic requirements of moving timeline structures around, and to offer as much as possible of what existing interchange mechanisms provide. We'll continue to add other aspects of AAF support as the specification evolves."

AAF support will also be available soon to Final Cut Pro users, though not directly from Apple. Instead, AAF will be among the capabilities supported via plug-ins that utilize a new XML interchange format incorporated into FCP 4. "It's is a truly open specification," Apple's Saccone says, "encoded using industry-standard XML. The format is a superset of AAF, and encompasses everything Final Cut Pro 4 can do. The implementation provides complete access to all project attributes, including media, assets, filters, transitions, edits, color correction, and keyframes." The format is being used by Automatic Duck to develop both AAF and OMF import/export components for use with FCP 4.

Additional forays into the AAF field have been reported on the part of companies such as BBC Technology, Da Vinci, Nucoda, and Quantel. But other vendors, including some who are members of the AAF Association, have yet to reveal the specifics of their AAF plans. "Pinnacle Systems has not yet made any announcements regarding AAF support in its products," Schormann says. As for Adobe, David Trescott, senior director of digital video, says the company is committed to providing AAF interoperability in future versions of both Adobe Premiere and Adobe After Effects, but "cannot discuss the dates or features of unannounced products."

Discreet takes a similar stance. "We are looking at all our products," Patel says, "but it is not our policy to comment on specifics of features until 90 days before we ship." How soon that might be, he says, depends on "when AAF makes sense as a richer means of data exchange than EDLs or OMF. Recent developments have been promising, and we feel that we are at a turning point where other vendors' AAF capabilities are becoming more meaningful. But we still feel it will be a while before cross-vendor AAF data exchange will be more meaningful than EDLs and OMF."

Cross-Vendor Commitment
Maurice Patel's "cross-vendor" qualifier underscores a key point, which is the issue of how committed vendors are to interoperability beyond their own brands. "We are using AAF for interchange between Avid products," Claman says, "but we are also actively working with other vendors to make AAF a valuable mechanism for metadata exchange between disparate systems. We now have several AAF implementations from different manufacturers on different platforms, but as with any new technology, a certain amount of work is needed to get different implementations working well together. I expect that it will take a few months to define compliance, after which—perhaps six to nine months from now—we'll see the first wave of officially ‘AAF-compliant' products."

Tim Claman adds that AAF "now has traction among manufacturers," and he's not alone in his upbeat assessment. "If the level of AAF acceptance and commitments is any indication of how much effort is put into adding AAF support to products, then the adoption rate will continue to be rapid," Media 100's Keilty says. But he adds that a difficult economic environment will force manufacturers to prioritize development, and that customers will need to continue to demand AAF support.
Leitch Technology's Nann, meanwhile, says that AAF already has significant momentum. "It's appearing in an increasing number of post-production tools, and many of the larger customers for these tools are making it a purchasing requirement," he says. "I expect that we'll see it as a standard industry-wide feature within the next year, but with varying levels of implementation. Then there will be a transition period as workflows, pipelines, and even user mindsets are adapted from the existing interchange mechanisms. And even once that happens, it will probably be a bit longer still before AAF lives up to its full potential."

The ultimate boundaries of that potential are yet to be fully explored, but Discreet's Patel urges caution in forecasting the format's impact. "It is important to be realistic," he says, "about how different manufacturers compete and collaborate. The situation where the tools of any vendor are freely interoperable with those of any other vendor is a utopia that will never be achieved."

The most likely scenario, he believes, is one in which vendors utilize private AAF data to guard the specialized capabilities that differentiate their products from the competition, while supporting public metadata interchange in "core" areas that are already informally standardized across applications. "The power of AAF," he says, "is its ability to adapt and expand with time to offer ever-greater levels of interoperability. As each vendor moves on to new generations of competitive tools, the pool of core data that can be shared will increase."

Companies Mentioned in This Article
AAF Association www.aafassociation.org
Adobe Systems, Inc. www.adobe.com
Apple Computer, Inc. www.apple.com
Autodesk, Inc. (Discreet) www.discreet.com
Automatic Duck, Inc. www.automaticduck.com
Avid Technology, Inc. www.avid.com
Leitch Technology Corporation www.leitch.com
Media 100 Inc. www.Media100.com
Pinnacle Systems, Inc. www.pinnaclesys.com