Multimedia
Blogs
Forum
Affordable HD
Whitepapers
Newsletters
Events
Advertisers
Blogcast
MillimeterNew Wave
Sep 1, 2001 12:00 PM, by Bob Turner
Web-Expanded Version
 The French film Vidocq was edited on Quantel iQ, which is fully AAF compliant; Quantel is a founding member of the AAF Association. |
If I told you a story about someone who was still using 30-year-old television technology in postproduction today, would your first reaction be a smirk?
Well, if you ever use more than one computer platform for postproduction, or software from multiple manufacturers, or a second platform for compositing, or if you import or export an EDL, then you may be using such a technology yourself.
CMX invented the EDL more than 30 years ago. CMX went out of business more than four years ago, having outlived the need for most of its technology. Yet many of us still use this ancient CMX EDL technology as a means of sharing editing decisions. I find this a disgrace in this digital, nonlinear era. To make matters worse, there is a solution, but the manufacturers need to know that their customers would like them to implement that solution.
I am referring to the Advanced Authoring Format, usually referred to as AAF. AAF is a technology specifically designed for digital nonlinear postproduction and authoring. It is a “SuperEDL,” but it is also much more. Its purpose is to solve the problems associated with the cross-platform interoperability of different manufacturers' postproduction and authoring software.
|
Click here for |
Now we have a hodge-podge of formats that go into and out of the postproduction process. Furthermore, the media can make more “roundtrips” than expected when the EDL was first developed. Today, these many media processes could include versioning, repurposing original content, outputting multiple formats to multiple delivery media, and workgroup processes, such as audio sweetening, graphics/effects compositing, offline, and finishing.
There is a growing need for a solution that allows work to cross over multiple platforms, operating systems, and various software from different manufacturers without the loss of creative decisions made on any one software or platform. Users want to eliminate the tedious redundancies of recreating such creative decisions as audio levels, custom effects, and titling. Users want to avoid the lose-lose choice between re-inputting original media or performing a time-consuming transcoding process because the media that exists is incompatible with the process you want to use.
More importantly, users want more than an EDL that is limited to one video track and four audio tracks in a world of many layers of video and audio. We want an exchange interface that deals with issues of multiple compression schemes, multilayer composites, multiple versions and revisions, and multiple format outputs with multiple aspect ratios. We want EDLs to include:
-
Library information about where a tape is shelved, where a file is archived, where certain files are on a network, or the original sources to a complex composite. Users no longer want a graphic that says “media offline.” We want the software to find the clip and bring it into our postproduction platform.
-
Better audiovisual clip identification. How many sources on today's EDLs are listed with the same AUX identifier? How many times do users see and edit and not know what the clip is? How many times do users have to retype descriptions?
-
Media asset management information, including the tools for find/sort/sift by multiple parameters and access to all the categories of a postproduction bin.
-
Scheduling information.
-
Effects information to allow transitions, plug-in effects, composites, and 3D DVE effects to be recreated automatically. Users also want to be able to recreate color processes and audio adjustments — including EQ — automatically on a different software package or platform.
-
Information to facilitate interactive authoring.
-
Ease of working in various frame rates cross-platform, including but not limited to 24p.
AAF technology solves such problems in a number of ways through the use of metadata. There are several different dictionaries of metadata, used for different purposes. The broadcast scheduler, the assistant cinema cameraman, the librarian, defense intelligence, and the Web-based interactive datacaster all have different metadata requirements. These sets of definitions are collected and documented by the Society of Motion Picture and Television Engineers (SMPTE) and coordinated with the European Broadcast Union (EBU). AAF is fully compliant with the EBU/SMPTE Metadata Task Force metadata architecture. This task force oversees and documents the development of metadata types and assists with design of software architecture that will provide cross-manufacturer compatibility for these different types.
Postproduction file interchange
The marketing director of a leading compositing software manufacturer recently acknowledged that AAF is to digital editing/compositing what the EDL was to linear editing.
|
Click here for |
The AAF solution has several components. It supports the complex combination of a piece of essence and its associated metadata, allowing the content to be described as a media object. AAF facilitates the cross-platform interchange of metadata and/or program content, and it is designed to replace OMF. It allows operators to track the history of pieces of program content from source elements through final production, and it permits access to networked content files on remote platforms or storage. AAF also allows the combined essence in a selected format to be rendered later in the postproduction process. It catalogs an extensive list of audio and video effects with a rich set of built-in standard effects. It provides a convenient way to wrap all elements of a project together for archiving. AAF encapsulates HTML and XML content, includes translators, and supports HTML/XML output, and it provides for Application Program Interfaces (APIs). Finally, although designed to eliminate digital islands of incompatibility and redundant tasks to expedite project management, AAF also allows a manufacturer to include proprietary metadata.
The inclusion of proprietary metadata allows the manufacturer of a product with advanced features to have that metadata included. Products created by that manufacturer are the only ones that can access it. An example might be a feature like Avid's script notation alternative to bins-base media management. At present, Avid holds a semi-exclusive license for this technology and could include proprietary metadata in the AAF without Discreet, Quantel, Media 100, Adobe, or FAST products being able to access that proprietary metadata. The other manufacturers can add their own proprietary metadata without Avid's product being able to access it. Thus the format allows for a rich compilation of standardized metadata and a portion of proprietary metadata.
Free open-source standard
Development of AAF is the responsibility of the AAF Association, incorporated February 2000 as a nonprofit organization (www.aafassociation.org.). The association is made up of a variety of hardware and software manufacturers and end users who want to advance media and metadata interchange.
The association's original board of directors included representatives from AVID, the BBC, CNN, Discreet Logic, 4MC, Matrox, Microsoft, Pinnacle Systems, Quantel, Sony, Turner Entertainment Networks, and the U.S. National Imaging and Mapping Agency. Today, members include: AIST, Avid, the BBC, CNN, Discreet, Dutch Broadcasting Services Corporation (NOB), EMC, eMotion, Encoda Systems, Fox News Corp., Front Porch Digital, Grass Valley Group, Informix/Ascential Software, Leitch, Liberty Livewire, Matrox, Microsoft, National Imagery and Mapping Agency, NL Technology, LLC, Omneon, Panasonic, Pandora, Philips, Pinnacle Systems, Post Impressions, Quantel, Sony, Tecmath, Turner Entertainment Networks, and Warner Bros. The wide range of members means that the standard is designed by industry-wide consensus.
It was announced at NAB 2001 that the board of directors voted to make the technology open source. (LINUX is another example of an open source solution.) This means that the software development kit (SDK) is free and downloadable from the website. Version 1.0 of AAF has been available since that announcement. Version 1.5 is expected at IBC 2001, and version 2.0 is hoped for at NAB 2002.
You may wonder why anyone would want to be a member of the association if the software is free. The answer includes opportunities to shape the future evolution of the software architecture and prioritize new features, as well as benefit from the association's publications and developer technical support.
In addition to its planned compliance with umbrella metadata standards under the supervision of SMPTE and EBU, the AAF Association is working with the International Standards Organization (ISO) and has a close development relationship with the Pro-MPEG Forum. The Material eXchange Format (MXF) is a fully compatible subset designed for simpler editing (e.g., news) and server and streaming needs.
Also, at NAB 2001 there was an AAF/Pro-MPEG booth display where several manufacturers' systems displayed the benefits of such interchange technologies. It was impressive.
Essence media
Although AAF was primarily designed to deal with metadata either married to essence media or linked to essence media but located elsewhere, AAF can be a native essence media file format as well. Such a file format, with essence wrapped in metadata, is mainly designed for postproduction needs. Remember that the ancestry of this technology includes the OMF interchange format. QuickTime is focused on information needed to play or stream the file. The AAF technology has a different focus. It is married to metadata that can facilitate postproduction processes with a history of processes from the original creation or storyboard through the final editing/authoring. Such information can make it easier to repurpose content or provide distribution capabilities through various media with differing aspect ratios and resolutions.
Call to action
AAF technology can improve the postproduction process. It is a far richer method of sharing editing decisions. It automatically provides metadata that is desirable for many aspects of postproduction. The technology can eliminate many tedious steps — saving time and simplifying project management. Valuable information normally lost when going from process to process can now be preserved. The ability to select the best software for each without data loss is now possible. For example, you do not have to select a finishing or compositing system from the same manufacturer as your editing systems simply because you want to preserve all the metadata. You can exchange media files without problematic and time-consuming transcoding steps or the need to re-ingest original source material. When incorporated into postproduction tools, AAF can save time and avoid brain-numbing tedious processes.
AAF products are available now. From hot, new, inexpensive AIST software to very expensive and incredibly powerful Quantel systems, AAF compliant products can be purchased today. If you understand and want the benefits of AAF technology, please contact your software manufacturers and ask when it will be available. Tell them you want their products to be AAF-compliant. Many manufacturers have said to me, “We are interested, but we just have not seen the demand yet. It will not be that hard to incorporate AAF technology when the time is right.”
The AAF Association has had a problem communicating the benefits of this complex technology to the average user. As soon as these users understand how it will save them time and money, provide better quality, and maybe even offer more creative choices, then I believe the demand will start to be felt. Don't let this be a “chicken and egg — which comes first?” scenario. Investigate what AAF could mean for you. Discover that the benefits I describe are only a part of what AAF will offer in the future. We can have a true cross-platform, cross-OS, cross-manufacturer transfer of content solution. We will be able to select the software solutions we want without compromise.
Data Dictionary
Essence data or essence media:
All program content consists of two parts, essence and metadata. Essence data is the actual audio/visual media. This data can be one or more of a variety of essence data types:
Video data: This can be uncompressed video, RGB, HDTV, 2,000 × 2,000 pixel moving images, proprietary video file architectures, or one of the common compressed video formats such as MPEG-2, QuickTime, .AVI, DV, etc.
Audio data: This could include any digital audio source (audio clip from a single track on an NLE, .AIFF, .WAV, Audio CD, DAT, etc.).
Graphics/CG/animation: This can be one of the many graphics, animation or text formats (.TIF, .TGA, .JPG, .PIX, etc.), plus you could have a proprietary or user defined essence, and a physical essence source such as videotape.
Metadata: Metadata is pretty much all the data that cannot be seen or heard. EDLs, timecode, and sync/blanking information is metadata. Today metadata can include:
Identification and location metadata: Anything that can identify a bit of essence media (e.g., file source type, location of a videotape, etc.).
Process metadata: This is how the essence media is assembled (how the media is composited or edited). This data may be used to automatically assemble a new essence media.
Administration metadata: This includes digital rights, encryption, user access, air date scheduling, and other business administration.
Interpretive metadata: This can be language descriptions, scene/take, character, camera angle, artists/creators, organizations, etc. Imagine all dialog automatically translated (speech-to-text) and associated and linked to the spoken word.
Parameters metadata: This could be signal coding, storage type, streaming parameters, etc.
Relational metadata: This includes the links, synchronization or relationships between essence media objects.
Spatial and temporal metadata: This could be creation dates, ingestion date, modification dates, durations, event times, delays, screen position of object, layer of object, etc.
History of EDL metadata
The CMX (CBS/Memorex/eXperimental) EDL established a proprietary list of edit decisions at the standard technology for transferring metadata or data that is not essence media (e.g., video, audio, graphics, text, animation, etc.) from a disk-based offline editing system (the Model 600) to an online finishing system (Model 200). This technology was developed in the 1970s.
The exchange of metadata (the Edit Decision List or EDL) took place by a paper tape. This was long thin paper with holes punched in it to record the EDL from the Model 600 and then input via a paper tape reader into the Model 200. The Model 200 videotape assembler product or “online finishing” editing system was followed by the CMX 300 in 1972 – it came with a keyboard terminal option for inputting and controlling the finishing process.
In 1973 came the unsuccessful CMX 50 editing product that was designed to work with U-matic tape. Two years later, the CMX 340 became the editing system of choice for postproduction. The CMX 340 and its successor, the CMX 3400, became the industry workhorses, and their popularity spawned competing products, including those from Sony, Paltex, and others. After CMX, one of the most popular early systems was the SuperEdit software of Dave Bargen. This worked first on a CMX system, then on the ISC (The Interactive Systems Company), which later was sold to Grass Valley Group and became the GVG SuperEdit system. Paper tape gave way to first an 8in. floppy disk, then a 5¼in. floppy, and then the 3 ½in. floppy. SMPTE tried to standardize the EDL with limited success, and the various manufacturers generally offered a choice of CMX, GVG, and Sony EDLs. The proprietary EDL included the in- and out-points for the source and record decks, split-edit information, and the type/duration of any transition. Soon there was a text line added to include notes the editor could add to the EDL, and a GPI (General Purpose Interface trigger command) was also included. In the 1980s, the list evolved to include “SloMo” instructions plus Switcher and TBC memory data. Later in the early 1990s came attempts to include DVE data, Closed Captioning data, Audio Snapshot instructions, and Read-Before Write capabilities for the record deck. Also, control interfaces to audio mixers, video switchers, DVEs, and DDRs were added. Generally, the EDL had one video “track,” and first one audio track, then two and then four audio tracks or channels to list timecodes.
Except for what an editor could describe in text, there was no metadata to describe resolutions or formats, compression codecs, revision information, or multilayer composites. The capabilities of the EDL did not include media asset management or library information, any form of scheduling, or complex 3D DVE instructions. If the editor typed in the wrong reel number (or forgot to change the reel number) there was no way of knowing. There was no automatic input of source information such as the name, creation date, or what was on the source tape. There was no method of automatically recognizing AUX sources, although in later versions the source could be named the same way a reel could be identified – by the editor typing it in.
The important point is that CMX developed this proprietary method of metadata sharing more than 30 years ago, and I wrote a published eulogy for CMX more than four years ago. It is time for an alternative for sharing metadata.
Two previous standards have attempted to create just that. The first was the Open Media Framework Interchange standard, sometimes referred to as OMF or OMFI. This standard had a rocky introduction and achieved only limited success. OMF was developed by Avid Technologies as an open standard, but competitors did not want anything to do with something that Avid developed. Some audio manufacturers, unthreatened by Avid and wanting better metadata from the dominant editing software manufacturer, found the concept much easier to accept. As this early success in the audio industry became understood, graphics, compositing, and animation software manufacturers took notice. Unfortunately, OMF was based upon an Apple Computer technology called Bentos, and Apple decided to abandon that technology in favor of something else. This appears to have killed further development – especially with the difficulty in gaining acceptance by other editing system manufacturers.
While OMF was struggling to gain industry acceptance, a number of editing technology manufacturers developing on the Windows platform also tried to create a set of standards to make cross-application cooperation easier. Open-DML was the name of this effort. Some interesting technology was the result, and a large group of manufacturers showed interest. But in the end this effort was abandoned.
Next came the trend toward workgroup postproduction, including separate composting workstations and more mix-to-pix work, which demanded better metadata-sharing than EDLs. Then as digital NLE editing began to dominate postproduction, the need for cross-platform exchange of both essence media and metadata became obvious. Most importantly, when companies saw that most offlines occurred on an Avid Media Composer and the reason purchasers chose an Avid system was because of the metadata exchange capabilities, it became obvious that something must be done. A group of leading software manufacturers (Adobe Systems, Avid Technology, Matrox, Microsoft Corp., Pinnacle Systems, Softimage, Sonic Foundry, and Truevision) decided to get together for a replacement for the OMF standard. This time the emphasis would be industry-wide support for the standard. Unfortunately, they started with a faux pas. On April 3, 1998, there was an announcement about a new specification that defined Media Authoring Interchange Standard. The misstep? It was announced by Microsoft at a time when the battle between Apple and Microsoft was at its worst. Needless to say, Apple has yet to support this standard. In spite of Apple Computer, the container format has been developed and industry-wide support has begun to take hold. On February 10, 2000, it was announced that the AAF Association was incorporated with a board of directors. The directors were representatives of companies such as AVID, the BBC, CNN, Discreet Logic, 4MC, Matrox, Microsoft, Pinnacle Systems, Quantel, Sony, Turner Entertainment Networks, and the U.S. National Imaging and Mapping Agency. A month later at NAB 2000, a large number of new members was announced.
Today, membership includes: AIST, Avid, the BBC, CNN, Discreet, Dutch Broadcasting Services Corporation (NOB), EMC, eMotion, Encoda Systems, Fox News Corp., Front Porch Digital, Grass Valley Group, Informix/Ascential Software, Leitch, Liberty Livewire, Matrox, Microsoft, National Imagery and Mapping Agency, NL Technology, LLC, Omneon, Panasonic, Pandora, Philips, Pinnacle Systems, Post Impressions, Quantel, Sony, Tecmath, Turner Entertainment Networks, and Warner Bros.
The AAF Association also has close relationships with and support from the SMPTE Digital Cinema Group, the Pro-MPEG Forum, the International Digital Cinema Festival, and GFORS. The AAF is committed to having its standard fully compliant with the work of the EBU/SMPTE Metadata Task Force.
At NAB 2001, working examples of AAF interchange benefits were displayed in a booth exhibit sponsored jointly by the AAF Association and the Pro-MPEG forum. The open source release of the AAF version 1.0 SDK was also announced.
Continue the discussion on “Crosstalk” the Millimeter Forum.