One of the meta trends we can take from this past year of COVID-19 is that content is still king. The end consumers want both quality content and they want it in a quantity that has, to date, been unprecedented. More people are watching more content on more devices than ever before, and the challenge has been to serve them not only with sheer volume but with relevant volume.
Recommendation engines have never been so vital as media organizations attempt to surface the right material at the right time to satisfy their customers. And, with advertising driven services in particular showing an appetite for older catalogue content to be re-watched and re-appreciated, the importance of library material is also vital. But one of the key issues facing media organizations is not necessarily always understanding what they have in their own libraries; effective Media Asset Management is only as good as the quality of the information in the catalogue.
We have seen an uptick in studios and content owners coming to us in recent months with their libraries and not really being sure what is in them. About a decade ago the mass movement towards digitization of library content started, but it was often done in a fairly chaotic manner. Everything got digitized, nothing got sorted, with the result that now media companies have what are essentially bug buckets full of files, many of which are duplicates.
Enter cloud-based storage for media libraries
And there can be a lot of duplication. When HD first appeared on the scene there were a huge amount of different sub formats created for it. 720p and 1080p are commonly known, but many broadcasters and streaming services added their own audio requirements to that, their own subtitling requirements, different watershed and censoring edits, their own graphics requirements, and more, with the result that it is easy for a video library to have a significant number of copies of what is essentially the same program.
Cloud-based storage has helped solve most of the pain points involved in running a large media library. Cloud-based content storage platforms provide rapid access, OpEx payments, and mean that companies can store their content more cheaply and access it more speedily than ever before. But that storage still costs and duplicates are a significant cost in themselves when you consider a large media library can easily run to thousands of titles.
This is where IMF comes in. IMF, the Interoperable Mastering Format, is an SMPTE standard that has been designed to store content in multiple formats with maximum efficiency. It works by creating a single master version of a piece of video content, whether it be a movie, a TV episode, or a commercial. Once that master has been made multiple deliverables can be derived from it by simply creating files that log the differences. These are stored as Composition Play Lists (CPLs) referencing the available essence components as required and only take up a handful of bytes, meaning that it is a quick and simple task to distribute differing copies of a title to, for example, a broadcaster, an online streaming platform with subtitles, and an airline that requires content edits.
Indeed, we’ve found that just by implementing it alone when working with a customer’s content library we can save companies around 50% on their ongoing storage costs as it removes the need for keeping entire duplicates of files.
Idiosyncratic abbreviations and 256-character filenames
As well as this tremendous increase in efficiency it also brings a degree of standardization to the way that metadata is described. Over the years this has been an ad hoc process to say the least. We know of major companies where the file naming protocols have changed five times in as many years, others where a single operator named files according to their own idiosyncratic abbreviations and the operator is now gone leaving behind 256 character filenames that defy any easy parsing.
With IMF you always know what you have and, crucially, so does everyone else that you deal with in the increasingly lengthy and distributed production chains that make up the modern media landscape.
In many ways its use echoes the way that other industries have developed, and Just-in-Time manufacturing in particular. In the same way that a car is no longer manufactured until the order is in and all the variables in the specification are decided, so a TV program or movie is no longer converted to a certain deliverable specification until it is ordered. With all the common elements already in place it is then a process of assembly, sometimes transcoding — using cloud processing we can transcode at 40x real-time — and delivery.
IMF with the cloud as an enabler
And mention of the cloud here is crucial as the cloud is the enabler of so much of this. With the cloud, everything gets pulled together into a single archive. Trying to manage tens of thousands of assets in a traditional storage environment, which can be spread across numerous different spinning disc formats, LTO tapes and off-line storage, is a headache at best. But with the cloud, everything is connected and everything is findable — the key is using something such as IMF to be able to locate it in the first place. We can even start using Artificial Intelligence and Machine Logic to sort through the content and apply the metadata tags for us, identifying language down to the dialect level, sentiment and more.
The speed of delivery and reduction in storage costs and footprint that IMF provides has already seen it becoming firmly established on the studios side of the business, but it can do more. It is not simply an output format, but one that productions can use at any point in the video production workflow
It is widely supported by many different vendors, allowing productions to choose different tools for different aspects of a job depending on what best suits at the time. There is no lock in to a specific brand of non-linear editor, for example; saving edits to the cloud as an IMF file means that most other software in the post process can simply pick it up and start working on it
And it supports the impressive degree of versioning that the increased use of addressable advertising is demanding. Targeted ads can segment audiences down to increasingly granular levels — we know of one pharmacy chain that needed to add in 2000 different pictures of their local store selected by geocode and pre-IMF that would have required 2000 separate versions of the ad. With IMF it is simply one master and 2000 separate CPLs containing the separate images, dramatically lowering the storage required and vastly simplifying the workflow.
Versioning and unique identifiers
Highly flexible versioning in film and post-production also becomes possible. Individual scenes, for example, can be given unique identifiers under IMF and then moved in and out of the supply chain for other facilities to work on. This not only makes project management an easier task, it also makes versioning for the sort of incremental changes that can have budget implications for completed productions (fixes or compliance edits, localizations, and so on) much easier.
But, back to the main thrust of this piece and its use in MAM. A good metaphor here is an Amazon warehouse. These giant buildings can contain literally millions and millions of pieces of inventory, but they know where every item is and how many of those items there are. Because of that it can pick, pack, and post inexpensive items to the consumer in incredibly quick time and still make a profit. And this is precisely what the use of the cloud and IMF allows broadcasters and other media companies to do; leverage their deep archive and provide consumers with what they want when they want it at a point where profit is still possible.