How to Choose the Right Format for OTT Video Library Management
By Matt Goble, Systems Project Manager, Visual Data Media Services
Having an efficient content supply chain is key to managing content across multiple platforms and formats, particularly in the cloud. But are businesses making the right choices when it comes to formats and video encoding? Not always.
The challenge facing media companies, especially in the library and archive space, is that there are no agreed-upon standards when it comes to both content production and content delivery. On the creation end of the value chain, productions are largely free to make their own choices, while at the distribution end there is a wide range of codec requirements, containers, preferred bitrates, color spaces, frame rates, and more.
This proliferation of formats has been driven mainly by the major VOD players as they aim to balance the demands of their end-users for high quality, non-buffering content, while retaining a degree of control over server and storage costs. As a result, the Library and Archive sector finds itself occupying the middleman position in these relationships and, thus, facing several challenges.
The prime issue occurs downstream, putting the onus on the library and archive teams to decipher each platform’s specifications and ascertain their requirements prior to delivery. While this information is often publicly available, it may be spread across many online sources, some that have been edited multiple times as specifications have changed over the years. And as platforms refine their processing flows—or, in the case of 4K and HDR, expand their offering—complexities increase and information becomes more disparate, harder to access, and far less reliable.
Meeting these specifications accurately the first time becomes an increasingly critical business objective for distribution providers. Platforms typically run exacting tests with new suppliers to ensure that the files being delivered meet requirements. And with more automation included in the ingest process, these tests ensure streamlined workflows and help avoid costs that can be incurred when specs aren’t met.
With so many options and reliance on spec compliance, how do media companies choose the right formats for their workflow? And how can they avoid racking up unnecessary costs that can come from duplicating storage or having to re-encode for delivery every time?
Why ProRes Fits Most Requirements
First, there is no “one size fits all” solution. But there is a “one size fits most,” a format that acts very much as a ‘universal solvent’ and can work in the vast majority of cases. If the content is SDR HD—and despite the hype around 4K HDR, most of it still is—then the most pragmatic option is Apple ProRes 422 HQ.
If we look at the current delivery specs for most of the major VOD platforms, we see that ProRes is both the most popular format for delivery and the format most widely supported. This is mainly because deliveries are handled by some degree of automation, and having a standard format reduces the likelihood of errors, as well as helping them balance costs to quality. Admittedly, some platforms, like Netflix, have recently unveiled alternate preferred specs, but even so they still accept ProRes. So, if the goal is to identify a starting point spec that gives the most flexibility in delivery and provides the most efficiency in the content supply chain, Apple ProRes 422 HQ is where to start.
If it’s that simple, why isn’t it the standard? Because it’s a propriety spec and historically the broadcast industry has supported open format with the understanding that it will offer the most flexibility. But, with the evolution of today’s workflows, ProRes has become ubiquitous enough that for most content owners it serves as a de facto standard for the distribution master. If a file complies with the Apple iTunes specification it can usually be delivered ‘as is’—creating more efficiency for the library involved. Also, if there is a requirement to transcode to another format, such as AVC Intra or XDCAM for broadcast delivery, starting the transcode process from a ProRes position provides precisely the level of quality required in the source file.
There are, of course, instances when ProRes is not the best solution. For example, when market dynamics alter the demand for 4K HDR, ProRes is not ideal. In this case, an IMF (Interoperable Master Format) thanks to its accommodation of higher resolution pictures, High Dynamic Range colour profiles, and immersive sound, offers the most flexibility. IMF also acts as a single master from which multiple deliverables can be made along with the requirements for each platform governed by CPLs (Composition Playlists). As we say, if the assets are the ingredients, then the CPL is the recipe. IMF is perhaps the ultimate expression of the ‘create once, publish many’ paradigm.
The Role of Metadata
Having knowledge of what those ingredients are is, of course, the role of metadata. Above and beyond any considerations of codecs, accurate metadata is vital to the successful operation of the modern media library. Without it we don’t know exactly what we have or what formats it’s in. Without clear and accurate metadata, the vital information required to meet platform requirements when it comes to audio, captioning, languages, rights and more, can’t be met with any level of accuracy or efficiency.
Unlike video formats, of which there are a select group to choose from, metadata schemas are proprietary to each company. Successfully building metadata, however, is less about adopting universal schemas, and more an exercise in following basic principles. Efficacy begins with ensuring at least all standard information like that mentioned above is present. New categories can evolve over time based on organizational needs, but setting the foundation will ultimately support the workflow long-term.
With more and more capable AI tools to help with the heavy lifting when it comes to metadata generation processes are becoming more efficient, though they typically still need a layer of real-time human-operator creation and validation. That said, AI tools can be great, and may offer useful ‘Optical Character Recognition’ for clocks and slates, as well as Logos and Internet Addresses. From an Archive standpoint, clients will often require Title Information to accurately identify content, Production Codes if they wish to hook into other products such as Rights Management or Ordering Systems, and then information about the Audio Tracks present in the master files—languages, audio layouts, track classes. All of this can be presented in a human readable form by AI routines, without having to decode a character-limited file name.
When it comes to choosing the right formats for library management, there are some clear benefits to choosing Apple ProRes 422 HQ for HD SDR content and IMF for the higher resolution—though much lower volume—4K HDR market. These both have good visibility into evolving demands and are future-proofed — at least for the foreseeable future. But it is important that they are coupled with efficiently generated, accurate metadata to give files the best chance of being accepted quickly and smoothly by the widest range of VOD platforms available, and ultimately enable the library to fulfil its position in the value chain between creation and production with minimal friction and optimized costs.