Signiant Moves Media … and Metadata
Signiant’s Metadata Everywhere series focuses on how our systems interact with and utilize the metadata associated with media assets. This shines a light on the control plane of our Software-Defined Content Exchange (SDCX) SaaS platform, which is where the system intelligence lies. Within the control plane, we process, store, and take action based on metadata from various sources, whether generated/extracted by Signiant or provided via other systems. This information can then be used to orchestrate the activities of the other half of the system, referred to as the data plane, which is where our advanced transport technology facilitates the movement of content files across the media supply chain. In simple terms, the control plane handles metadata while the data plane deals with media essence. The two parts of the architecture work together to maintain linkages between the heavy media assets and the information about those assets.
But there’s an interesting twist to this story as the metadata explosion continues within the media industry: sometimes our transport optimization technology plays a role in moving the metadata files themselves. In other words, it’s not just media essence that moves around the global supply chain via our platform’s data plane — Signiant products are also used to send and share large metadata files in a fast, secure, and reliable manner. No special adaptations are required for this use case. Our technology can move any data set, byte-for-byte, to and from any kind of storage anywhere in the world.
The Explosion of Data and Metadata in Media
In the increasingly competitive and fast-moving world of media, there are more and more kinds of metadata in active use. Metadata has long played a role in media creation, management, and distribution, and these functions are all becoming more granular and complex in the multi-platform, VFX-intensive world. In parallel, the business world at large is becoming more data driven. Media businesses are no exception, and with the increase in direct-to-consumer content delivery, they can now build data bridges to and from the people who view their content. All of this digital information about media assets is incredibly useful in optimizing both viewer experience and media business results, and modern technologies such as AI/ML can assist in deriving actionable decisions from the data. But, of course, it also creates new technical and organizational challenges. This recent press release from NBCUniversal illustrates the increasingly data-centric nature of the media business:
NBCUniversal Creates Chief Data Officer Role as It Streamlines Data Operations
The company named John Lee as its chief data officer as it builds out a central data function for all its units, including TV, theme parks and brand marketing.
Comcast Corp.’s NBCUniversal said Thursday it hired John Lee in the newly created role of chief data officer as the media company aims to bolster and streamline its data operations.
Mr. Lee, previously global chief corporate strategy officer at Dentsu Group Inc.’s data-marketing agency Merkle, will be tasked with overseeing NBCU’s data operations, including collection, storage and matching capabilities meant to help marketers target ads on the company’s media properties, as well as inform corporate efforts such as product development, said Krishan Bhatia, president and chief business officer for NBCU Advertising and Partnerships.
Mr. Lee, who reports to Mr. Bhatia, will also work to build a central data function to streamline efforts across the company’s divisions, such as its theme parks, TV and streaming business, film studio and marketing functions within its brands.
“The approach to data in most large companies has been one of executing at the business unit level, for specific use cases,” said Mr. Bhatia. “What NBCU is now doing is putting data front and center as an enterprise asset.”
The line can be blurry between data and metadata, but it’s not really a critical distinction for the purposes of this discussion. From a big-picture perspective, we’re talking about information about media assets rather than the content files themselves. The volume of this metadata is exploding and it is increasingly necessary to move it between various public cloud platforms and geographically distributed locations. At Signiant, we’ve noted that more and more metadata files are flowing through IP networks via our technology. These files can take many different forms, some of which are outlined below.
Embedded & Bundled Metadata
For completeness, we’ll start with the types of metadata most familiar within the media technology space. Technical and structural metadata is often embedded directly into the essence file, in alignment with a standard such as MXF. This type of metadata intrinsically travels around the world via Signiant technology, because that’s how its parent essence file is being moved.
Slightly more removed from the essence are separate text files, or “sidecars”, that contain metadata ⎯ typically using the XML format. These files usually contain contextual and descriptive metadata, captured either by humans or by automated systems. The files can become quite large when frame-by-frame metadata is involved. For the metadata to be useful as the asset moves through its life cycle, it’s obviously critical to maintain a direct linkage between the essence file and the sidecar(s) containing information about it. While this direct association between essence and metadata can be handled by various kinds of asset management systems, it’s often most pragmatic to just keep the two bundled together as they move through the supply chain. In lots of media workflows, the metadata sidecar files, therefore, travel right along with the essence via Signiant software. Whether or not the sidecar itself really crosses the threshold into requiring acceleration is beside the point ⎯ it needs to stay close to the essence and be transferred as part of a package.
Modern standards such as Interoperable Master Format (IMF) take the notion of an essence/metadata bundle a bit further. An IMF package includes various types of media essence, along with the metadata “recipes” required to assemble complete assets from the parts. These recipes, known as Composition PlayLists or CPLs, are XML files that reference various content components and define the timeline for a specific version of the asset. When you send an IMF package via Signiant, you’re sending both essence and metadata.
Content Creation Metadata
Many readers will be familiar with EDL and CDL concepts, but here’s a quick recap. An Edit Decision List, or EDL, is a metadata file format that allows information about edited sequences to be shared between applications and locations. By providing details about the source video files, the time in and out for each clip, and the associated effects, the EDL provides an efficient means of transferring timeline information.
A Color Decision List (CDL) is similar to an EDL, but rather than timeline details it contains data about rough color changes that have been applied to the footage. A standard defined by the American Society of Cinematographers (ASC) allows basic color grading information to be shared between devices and locations. In a typical use case, the CDL provides the mechanism for transferring an on-location “look” specified by the creative team to post production as a starting point for the final color correction. It’s important to note that the CDL captures only basic color adjustments, not the level of detail involved in secondary/final color correction.
In a modern VFX pipeline, EDLs and CDLs are often used to help facilitate the flow of information between the on-set DIT, the editing team, the VFX team, and the final DI or color grading facility. By efficiently propagating creative decisions through the stages, these metadata tools can enable highly efficient workflows in an environment that typically involves extremely high-resolution images and very demanding creative professionals.
At Signiant, we’ve observed that our customers in the production and post production segments are using our systems to move many different kinds of data during the creative process. Video and audio files are obvious, and, as discussed, the flow of EDLs and CDLs is prevalent in certain workflows. We also hear about productions using our systems to share digital files containing scripts, storyboards, studio notes, technical settings about cameras and lighting, scene descriptions, color Look Up Tables…and the list goes on and on. Not all file types require the heavy lifting of our accelerated transport, but security, visibility, and ease of use are always important. Our products are already in place to solve the problem of large file transport, so it’s easy to use these familiar tools for moving other kinds of digital assets that flow within the content creation environment.
Although few might describe it this way, modern media consumers expect to have a broad range of metadata-driven tools at their fingertips. They want comprehensive Electronic Program Guides, they like to search for movies starring their favorite actor, and they want to browse available content by genre. All of these capabilities depend upon access to a large pool of metadata associated with each media asset, if not each frame within that asset. Those metadata resources, combined with powerful search capabilities, are essential to today’s home video experience ⎯ whether delivered via cable, satellite, or OTT streaming. Consumer electronics companies have also entered the game with their own offerings built into television sets.
Particularly in the streaming realm, the flow of consumer-facing metadata has become a two-way street. Consumers expect their content providers to employ recommendation engines to predict what they might be interested in, and of course this isn’t possible until the system has collected information about each consumer’s viewing preferences. As that knowledge builds, the recommendation engine can correlate data about the viewer and data about media assets ⎯ so once again there are media metadata underpinnings that come into play.
Some providers of these consumer-facing services purchase the metadata from third parties, while others build their own proprietary databases. Either way, the data sets used for search and recommendation engines are huge. As consumer-facing metadata is bought, sold, created, and modified across the media ecosystem, Signiant software is often used for fast, secure sending and sharing of the files.
Business Decision Metadata
As illustrated in the NBCU press release cited earlier in this article, data relating to media assets plays an increasingly central role in business decision-making and revenue optimization within media enterprises. Targeted advertising, for example, is another area of focus for content providers that is closely linked with the consumer-facing applications noted above. If the media company knows enough about a consumer to recommend content, it’s a small step to also employ sophisticated targeted advertising that goes far beyond ZIP codes. This is not only a revenue generation opportunity, it also potentially enables greater consumer engagement and loyalty.
Content companies have long relied upon media consumption data to inform their business decisions. Box office receipts and Nielson ratings may seem crude by today’s standards, but they were nonetheless media metadata ⎯ information about content consumption. Modern media companies now employ very granular cross-platform audience measurement tools and services to understand who is watching what. The power of predictive analytics to make greenlight decisions remains to be seen, but content companies now have a lot of data at their fingertips to help inform business decisions. As these data sets become increasingly critical to media companies, the footprint for Signiant products often expands within the enterprise. Within some of our largest customers, media technology groups use our technology to send and share the content itself, while data scientists across the hall use the same toolsets for other business purposes.
In summary, the massive amount of data flowing throughout the global media ecosystem isn’t limited to the content itself. There’s also much more data about the content, and it’s used in diverse and interesting ways. The data explosion impacting every corner of society is also a factor in the Media & Entertainment industry, and Signiant’s tools have an important role to play.