Renard T. Jenkins, the public broadcaster’s senior director of operations: production, media and distribution, talks about why PBS is converting to the Interoperable Mastering Format for delivering programming to OTT services. While the move was spurred by a Netflix requirement, Jenkins says that because of the format’s many advantages “I think IMF makes sense. It is the right thing for television broadcasters to be looking at right now.”
PBS Prepares Itself For IMF Deliverables
File-based workflow specialist Archimedia Technology held a press conference Friday at IBC 2015 in Amsterdam, not focused on its new product introductions, but rather on IMF, the Interoperable Mastering Format, widely used in the film world but now beginning to gain traction in TV.
One driver helping to make IMF more attractive to some broadcasters is the decision of OTT giant Netflix to require its content providers — not simply movie studios, but also broadcasters — to deliver content as IMF files by the end of the year.
In the broadest of terms, there are two components to IMF files: the essence — including the video and audio content— and metadata. Together these two sides of the IMF coin make it simple for Netflix and others to extract the version of content they need to meet the individual tastes and needs of particular users or distribution channels.
Need an English-language version of a movie or TV show for an iPad with French subtitles? No problem. Simply mix and match the right IMF metadata components to the essence and voila, the desired version of the content can be created automatically — saving countless hours of work and money.
Archimedia put together a panel to discuss IMF and broadcasting. It included Stan Moot, CTO of the International Association of Broadcast Manufacturers; Pat Griffis, executive director technology strategy in the office of the CTO at Dolby Labs; and Renard T. Jenkins, senior director, operations: production, media and distribution at PBS.
Six months ago, PBS began preparing for the Netflix IMF delivery requirement, Jenkins said during the panel. When done, PBS, Netflix and the public broadcaster’s other OTT partners will have a much easier time delivering the 800 versions of Downton Abbey that Jenkins said must be prepared, as well as multiple versions of countless other PBS programs.
I walked with Jenkins to his appointment following the panel and asked him about IMF, PBS’s preparations and the benefits of implementing the Interoperable Mastering Format.
Six months ago, PBS got interested in IMF. Why?
It primarily was because one of our delivery points, Netflix, was really already going down that road and were giving us sort of a primer to let us know that they needed us to begin looking at what was their next delivery spec.
That is how we started going down the IMF road, saying: “Let’s investigate IMF, take a look at it and find out where it goes for us.”
That is when we started working with SMPTE and looking at all of their papers and trying to figure out how we were going to make this work.
What are the greatest challenges you have faced implementing IMF?
One of the biggest was the expense of deploying IMF. At the time we were looking, there were really two companies that sort of had a real working IMF workflow. Those companies were Telestream and Clipster, which was the more mature of the two platforms.
However, Clipster was very expensive, and for a public entity like PBS, we always have to pay attention to the dollar amount we are spending on something because we want to be aware of how we utilize the public’s money.
So when we started to look at Clipster, we said: “We have sort of an instance of Telestream. Can we get an upgraded version of Telestream that will allow us to begin to build?”
We were also working with Dalet and Amberfin, and they were going down the IMF road as well. But our primary platform to be able to test and do this is Telestream Vantage.
IMF is dependent on metadata. I am sure you were using metadata before the decision to move to IMF. Was that the metadata you needed to deploy IMF?
From our standpoint, the type of metadata we were using versus the type of metadata required in IMF, required a change.
We changed what type of metadata we were looking for. Like we described during the session, we are essentially going back in time. We are going back to an edit decision list. It just so happens that our edit decision list is being built in a file format and a file workflow.
So for us, we had to stop and say: “OK, what is it that we want to get out of a particular file? How many different iterations do we want from this particular file and how do we make that happen?”
So [we] began to look at the type of metadata used. We are now looking at metadata from the standpoint that we know we need to have a cut at timecode XYZ because we know that this is going out for G-rated regional playout instead of a PG.
In that sense, we actually began looking at how we create our metadata and the things that are important to us. And those are the things that we are experimenting with now.
Does that mean that you had to go back for all of your existing programming and create all of this metadata for everything that’s on the shelf or in the server? Or, is it being done from some specific date and going forward?
I think the main thing for us is going to be what’s important to our public. What shows and programs are we going to be required to get to our distribution point? That’s really what is going to dictate what it is that we do, how often we do it and how much of this metadata creation that we do.
But in the background, we are constantly looking at our library because we have long-tail content.
So, we will look at that and likely have a background process going with metadata being generated from that, and right now we are in the process at PBS of defining what the landscape is going to look like going forward because we have tons of tapes that have existed for years, but the metadata on those tapes is based on the log that’s inside those tapes, what label is on that tape.
We now are defining what our metadata landscape looks like. We are now making sure that as we continue to work towards where we are going to go with our metadata that we are making every effort possible to cut down on the amount of extra work that we are going to have to do.
So it was Netflix that prompted all of this?
Yeah, what prompted our decision to move into IMF and this type of work was Netflix saying this is what we will require in a certain amount of time.
In doing that we decided to look at this; however, we also have other things going on in our organization as we look at our network playout system.
I am in the midst of looking at solutions right now to get some of our legacy systems into a more IP-based environment that will require a very file-based delivery to our environment as well as what we are spitting back out.
Those are the things that are driving what we are doing constantly at PBS. So this sort of dropped in at the right time for us to continue doing that.
If you were not a public broadcaster, would deploying IMF as a foundation for metadata make sense? In other words, if you didn’t have to worry about delivering to Netflix or some other OTT third-party that needs to offer different versions of the essence of the content, would IMF make sense?
Yes. I strongly believe it would make sense because even if you are not deploying to Netflix and other entities, the logical way in which IMF is created just makes plain sense. It really does.
The way I look at it is as either a well-constructed suitcase for travel around the world, or as something like the Food Pyramid. You’ve got every single food category represented on the plate. What you decide to eat is up to you.
Personally, I think IMF makes sense. It is the right thing for television broadcasters to be looking at right now.
So you have all of this IMF metadata to deal with. That must pose some quality control issues you didn’t have to deal with before.
Yes it does. You actually have to 1) be able to teach people how to read XML because you can use XML in a human-readable form to see what it is that you actually are creating; and 2) you need to have a situation where you can check the file properly.
QC-ing an IMF package is not like running it through a file check. It actually involves opening up the package, making sure all of the components are there and that all of the components are true in the sense that I may have my OPL (output playlist), but is it the right list for the destination to which it is going?
The joy of this is you can have multiple output lists point to a single CPL (composition playlist) and you can have multiple composition playlists pointing to a single output list.
If you are going to go down the road of having a single output list, it has to be right. That’s where human-readable XML enters, and it is where, really and truly, the best QC tool you could ever have is an educated human being.
That’s where we are right now.
Our first delivery of IMF was interesting. We got some things right, and we had other things that were just wrong, even though we had thought we were going down the road and checking off all of the boxes.
It’s a learning experience, and QC is very important with IMF, just like it is with any file-based workflow.
Is there anything else you would like to add?
To me the biggest things about IMF is it is a way for us to create the hundreds of versions [of a program] that are needed, and a way for us to give ourselves flexibility in a situation where we are generally looking for reliability over flexibility.
We sort of have built a system for PBS right now that is very much structured on being able to deliver 99.99% of the time to our member stations and to our viewers and making sure that we are giving them the best that we can give them. So we have cut out flexibility in doing that.
IMF is giving us the opportunity to return to that flexible state where we can deliver that 99.99% of the time, but also have the ability to say: “You can have all of these different flavors.”
We are not going to have to increase what we are doing in order to deliver that. We can now open the door to doing even more thanks to IMF [since it] allows the end user to determine which output playlist they want to use.
That for me is really important. But remember, for us, this remains in the realm of OTT deliveries. It remains in the realm of our digital distribution points like Netflix, Amazon and so forth.
We are not looking at it at this point for our real-time delivery of content to our member stations. Any change to that, anything we do in that realm, would be a much larger lift.
To see all of TVNewsCheck’s IBC 2015 coverage, click here.