AMWA ratifies MXF Versioning specification
The Advanced Media Workflow Association (AMWA) has released Application Specification AS-02-MXF Versioning. The specification is a constrained MXF file format that meets the needs of content creators and distributors who are looking to improve the efficiencies of delivery to the multiple platforms that now proliferate, as well as meeting the multiple languages needed for global distribution.
“In a nutshell,” said AMWA executive director Brad Gilmer, “this AS allows a broadcaster to collect all of the different components used to create many different versions of the same content (a movie, for example) in one place and then render out a particular version (the movie in English with Closed Captioning, Italian subtitles and a Spanish audio track, cut to fit in a two-hour timeslot).”
Ever since video and audio content began being handled as digital data, there has been a quest for simple file exchange between equipment and software applications from different vendors. The SDI interface for exchange realtime streams set the initial benchmark for interoperability, and broadcasters aspired to a similar simplicity for file exchange.
Although containers like QuickTime and AVI provided formats for file interchange, they did not meet many of the special requirements of the broadcast and film sector. This led to the formation of a working group to formulate a set of standards that culminated in the SMPTE MXF standards (377M-2004 and others).
The MXF file format is very flexible, and, to aid its adoption, a number of Operation Patterns (OP) were defined which constrained MXF. OP1a (SMPTE 378M) is the most used, and specifies a single item, single package. It is analogous to a videotape in that it can contain interleaved video and audio tracks, and can be used to interchange clips or program items between television equipment.
An example of OP1a in use is the file container used by Sony XDCAM products record to the Professional Disc, and for delivery of files from editing systems to playout servers.
A specialized Operation Pattern, the OP-Atom has also proved popular. This defines a single essence item, like a video track or audio track. An external mechanism is used to synchronize a group of such items.
Panasonic camcorders that record to P2 cards use the OP-Atom specification as containers for the individual video and audio tracks. OP-Atom can also be found at the heart of Avid editors.
As the use of MXF spread, it soon became apparent that the standard in its wide scope and flexibility was proving to be an impediment to interoperability between different vendors’ broadcast equipment. Although equipment from one vendor generally interoperates, exchanging files between vendors’ equipment proved to be more of a problem.
A simple example could be exchanging an OP1a file. The source equipment encodes as D10, MPEG-2, but the receiving equipment only supports AVC decoding. There can be many more obscure incompatibilities between equipment that, apart, can conform to MXF standards. Mutually, however, that is not the case.
Broadcasters like Turner had expressed a need for a file container that would facilitate the management of different version of program items for different platforms and different devices. Broadcasters in general, are looking at more efficient ways to create version to meet the needs of long tail delivery, where revenues are much lower and current cost-bases prohibit the profitable exploitation of the long tail.
It was these problems of interoperability, and the complexity of file-based operations that led the AMWA to try to address the many issues. The AMWA had had great success with the AAF, used in post-production for project interchange between for example video editing and sound dubbing. The AMWA is driven by real business needs, with technology being the servant to those needs.
The AMWA working group proposed solving the problem by adding further constraints to MXF, beyond the OPs. This leads to a simpler specification, which easier for vendors to comply and support.
The specifications are designed for one application in the workflow, hence Application Specification (AS). In the case of AS-02, it is to create an MXF master file from which distribution versions can be simply created. AS-02 MXF Versioning is designed to serve the needs of broadcasters and content creators that need to create many versions of media to suite the multiple formats needed for the many devices now used to view content. The International nature of the media business also requires the creation of many tens of versions of programming in different languages and for different regulatory regimes.
“Part of the reason for AS-02 is to put in place a specification that mimics the commands coming down from the command and control structure that drives the workflows,” Amberfin CTO Bruce Devlin said. “For example, ‘fix the loudness on that audio.’ Why are you streaming all the video to fix the audio?”
Traditionally, this meant many tape versions carrying different edits and different language tracks, along with all the different captioning and subtitling needs. Clearly this process could be costly and inefficient, and, for global media companies, an asset management challenge. The advent of OTT video, tablets and mobile video has only increased the problems of managing multiple versions.
Application specifications do not replace Operational Patterns (OP). Instead, they use all the many aspects of the SMPTE MXF standard and fully comply with them. The AS may contain OPs nested internally. For example, a video track (picture essence) can be OP1a wrapped within the media folder of the AS-02 container.
AS-02 MXF Versioning
AS-02 is a container that wraps all the components that can be used to assemble a version of the program, as well as any other files the broadcaster may wish to wrap with the audio-visual essence. The AS-02 specification adds several features over simpler MXF files like OP1a in order to aid the processing of media files. These include a file manifest, shims, and an extras folder.
AS-02 stores all the essence components in the media folder (see Figure 2). Essence tracks are stored as separate MXF files. Interleaving is not supported, as that removes the flexibility to simply create versions. MXF supports popular video essence like D10, AVC and JPEG2000 (many implementations of OP1a, especially for streaming have interleaved video and audio).
PCM audio files can be mono, stereo or surround, and can be considered as a single track. Different languages should be stored in separate files.
The media folder can also store closed caption data and subtitling data carried in VBI or VANC, and not interleaved with the video. Again, by not interleaving, this facilitates post-processing.
AS-02 playback devices use the version file, then the index files in the essence files, to render essence in sync.
AS-02 supports a Media Integrity Check (MIC) for essence components. This has been requested by broadcasters, as tests have shown that with a very small proportion of file transfers there can be file corruption. The MIC allows this to be instantly flagged for remedial action.
The heart of the AS-02 bundle is the version file. This is a single MXF file and it references the essence components, which are stored in the media folder. A simple version file is OP1b. Segmented or editing bundle use OP2b and OP3b respectively for the version file.
The root folder of the AS-02 bundle contains an XML manifest file with a list of all of the files and folders in the bundle — a “packing list.”
Every broadcaster has their favored codecs, possibly driven by camera choice, but also to suite their workflows. The “shim” is a further constraint to an AS that can define picture resolution, codec, bit rate, audio coding and other parameters. For example a broadcaster may want to use D10, a film studio may want to use JPEG2000. The shim can also define parameters like audio track layouts, surround and stereo.
This folder can be used to store non-MXF data, which could be metadata or general text files. Some examples include subtitle files, Q.C. reports, loudness control data and scripts (see Figure 3).
Non MXF-aware applications
As an AS-02 file is a regular folder, with nested folders and files. Desktop applications can open the folder and read them, for example a quality report as an MS Office document. Such functionality could be integrated into a DAM system.
Earlier implementation of media file wrappers may well have needed a non-linear editor to be used to read such embedded data. AS-02 represents a step forward in flexibility for the general handling of media files and associated information necessary to operate a media business. Ian Baker of Metaglue explained, “ MXF has come of age. It can be used as a toolset to achieve business goals, rather than just a file interchange format.”
Much of the focus of AS02 has been on broadcast application. In parallel the film studios have been developing the Interoperable Master Format (IMF) with similar application to AS-02. It has been developed by the Entertainment Technology Center (ETC), and uses AS-02 media wrapping. The version file is XML rather than MXF.
To illustrate the advantages of AS-02, consider the creation of a new language version of a program item. A proxy is sent out to dubbing and subtitling facilities, which create the necessary WAV and subtitle files.
To add the audio, the file is added to the media folder and the version and manifest files updated accordingly. This involves very small file transfers. Traditional methods would have required the entire folder, with the large video files to moved across the network, just adding to network traffic and loading.
Similarly, VANC essence can be updated without touching the video. It is expected that new applications will develop as AS-02 comes into common use in broadcast facilities.
AS-02 is a format, not just for versioning, but could be used as the master file format for a program. It adheres to the SMPTE standards for MXF, yet is constrained in a manner that provides for interoperability, avoiding some of the inevitable vendor lock-in of earlier file-based media systems. It includes:
- a version file referencing the essence files
- a manifest
- extras folder, for non-MXF data
- shims to constrain codecs
- vendor neutral
AS-02 meets the needs of the modern “media factory” that needs to generate multiple versions, and multiple languages, for delivery to multiple platforms. Audio and subtitle layback becomes much simplified.
As a master folder, it facilitates the movement of content around the world — essential for global media companies. The format is starting to see support from vendors, and, with the final release, one should expect to see further support over the coming months.
The full specification can be downloaded from the AMWA web site.