 Live from Las Vegas, it's theCUBE, covering NAB 2017, brought to you by HGST. Hey welcome back everybody, Jeff Frick here with theCUBE. We're back at NAB 2017 with 100,000 of our favorite friends doing everything about broadcast media. It's media, it's entertainment, it's technology, it's the Met Effect, which is all the rage here at the show because you can't really separate the three. They're all tied together and really excited to be joined by our next guest in the weeds, keeping an eye on this, trying to keep up with all the crazy trends. He's Avi Swardlow, he's a manager, research and development at the Walt Disney Company. Avi, welcome. Thank you, thank you for having me. Absolutely, so first off, we talked a little bit before we went live your first time of the show, kind of general impressions of NAB. Yeah, it's big, a lot of walking, it's my first impression. Aside from the tired feet, it's really exciting to see all the new tech out here. From talking to other people who have been in years past, it seems like things move really fast here. So what you were seeing last year was completely different from what you were seeing this year. But loving all the different sections and everything from hard work to kind of some of the more data-driven stuff. Noticing that a lot more things are moving digital, that a lot of demos are now on laptops instead of physical, which is exciting to see. Have been impressed by some of the bigger company like Microsoft and IBM's machine learning efforts, and equally impressed by some of the hardware plays at DJI and GoPro. So really, really exciting stuff. It's really interesting kind of bifurcation of the market. On one hand, you've got all this crazy high-end stuff with 4K and 6K and 8K and Ultra HD and all these things and 360 and all these crazy cameras. At the other hand, you've got this democratization of distribution with YouTube and Vimeo and all these tools being brought down in a price front, you know, Samsung 360 camera, where you can be a relatively small content creator and have amazing tools at your disposal. So the opportunities from a creative point of view have probably never been richer. Absolutely, and I think a lot of what we're trying to focus on is moving in that digital direction for some of our content, trying to implement some of those lower-end or more cost-efficient tools and those distribution points to get our content to people faster, while at the same time trying to keep up on kind of the higher 4K end. You know, something that's interesting when chatting with my colleagues is that things move so fast that it's hard year to year to come here and see all the new things that, you know, are completely different from what you saw last year and yeah, you have to start implementing those things. So I think it's a balance between all of that. I think, you know, given that we're a big media company, some of those lower-end tools are really interesting to us in the sense that, you know, take news, for example, it's equally exciting to go live on Facebook video as it is sometimes to do it on a traditional broadcast. So I think learning how we integrate those and integrate those well are some of what we're trying to explore. Right. One of the topics we talked about before the cameras turned on was kind of this virtual reality and augmented reality, VR and AR. And it's pretty interesting as you talk specifically about, you know, data infusion kind of on top of techs. And I remember the first time I ever saw a sports broadcast where, I think it was Fox, maybe that put the score bug on the upper left-hand corner, and like, what is that? You know, you're taking valuable real estate. Now we're so accustomed to these multilayers of data on top of their broadcasts. You take like a Bloomberg channel where some of those things, where now they have multiple feeds that are constantly going. It's a very different way to consume data, but that's what people really want these days. Absolutely. I think that, you know, last year was kind of this year of AR, VR. People thought there was going to be this massive revolution all of a sudden where everybody would have headsets and VR would become ubiquitous. And I think that will happen eventually, but it's probably going to be a slower burn, mostly because people don't have devices yet. And I think there's not enough content out there, not enough devices out there. Regardless, I think that if you're still down what AR and VR is at its core, it's the augmentation of information over something else. Right. And so I think, you know, a lot of people are now starting to explore what are the baby steps you take to implement some of that technology into your workflow, assuming that people don't have devices yet. So I think, you know, when I look at some of the virtual sets that we're seeing around the show and the implementation of information over, let's say news or sports broadcasts, that becomes really interesting. And if you can use, you know, we're talking about photogrammetry or volume capture. If you can use some of that and do interesting stuff, for instance, like if you're looking at a sports game and you're able to create, you know, in something like Unity or Unreal, an asset that represents the sports game, it becomes a much easier way to understand what's going on in the game than just a set of numbers. So it's, you know, yes, when you saw that score in the top left-hand corner, that was exciting. Now imagine seeing a live 3D version of the game. Same information unfolding just in a different way. Right. And so I think those are the baby steps towards this AR, VR implementation. And eventually you might get to a point where everybody has a headset, but baby steps to the average consumer. Right, right. And a lot of conversations about machine learning, and you said, you're excited about some of the machine learning. You've got kind of just the metadata and better metadata around the assets themselves, but now actually getting into the assets at the frame level to do more exploration so that people can, you know, see age old adage, right? Find, consume, and share the stuff that they're most interested in. So there's a lot of new opportunities because of the horsepower of these machines here that we're surrounded by in terms of the massive capacity and speed of the storage systems to do things that you really couldn't do inside the assets themselves. Absolutely, and I think our problem at somewhere like Disney is unique. It's different than a Google or Facebook. We're not looking at this huge well of content like YouTube. We're looking at a smaller amount of content, and what's really important to us is accurate metadata about our content, more so than just having metadata. So a lot of what we focus on is definitely metadata extraction, but to the extent that we're going to use these machine learning tools, we want to have really good training sets and get back really accurate data. So a lot of what we focus on is being able to have a QA layer on top of the machine learning efforts, being able to use machine learning efforts that can be honed towards one show, for instance. So we're only extracting a certain set of characters. So we really enjoy using these tools and enjoy finding ways that we can apply them to our unique problem, which seems to be different than the problem that some of them are trying to address. But regardless, they're working really well for us. So what are some of the use cases, or can you share any of how you're using machine learning to get and score that kind of metadata? Yeah, for instance, we're starting to use metadata in some of the ways other people are, and some of the stuff that I can talk about, for instance, is facial capture, location capture, things that other people are doing, but again, they're unique to one show. For instance, a Quantico on ABC might be something where we have a set of characters that we're looking for. We're starting to use machine learning to look at things like that. Interesting. Now, Disney, obviously, great company, been around forever, huge legacy. Just curious the conversations in the hallway as there's this crazy wave of technology kind of budding up against, we still have to tell great stories, and Disney has a long history of telling great stories whether through the original animation studios or all the vast properties of which you guys have grown up. Is there still kind of a creative kind of ying and yang there? Is there a threat and kind of a rebalancing about kind of technology versus, let's not forget what should be job one? Absolutely, and I think that's why I really enjoy working at Disney. It's always story first, and my background is actually in creative development in the film industry, so I always come at it from a story first point of view, and I enjoy that the rest of the company does as well, but if you look at kind of Disney's history, it's always been technology complementing story. Think about the multi-plane camera and Snow White. The reason Snow White was able to be made was because Disney democratized animation. He figured out the technology that made animation possible at a feature film scale. Without that machine, that would not have been possible. So I think in our core history, you have these certain technologies that are put to use in the service of story, and I think that's pretty much how we approach everything. We're looking for stuff that's going to augment our storytelling efforts, not replace it, not degrade it anyway, but only to enhance it, and that's in our legacy. Right, right, that's interesting. I've never heard it explained that way, but that is so much the trend that we continue to be on today. It's democratization of the data, democratization of the access to the data, democratization of the analytics of the data, and then operating at scale, which requires that today's scale, we're not talking about two-hour movie scale and actually being able to produce that animation, but just massive amounts of data that are flowing through the system. So how do you operate there? We want to use that data to empower our storytellers, to empower anybody at the company to tell better stories. But the data management, it's tough. I think a lot of what we had to do is, first of all, put in place the plumbing to make that data easily accessible, to make it easily searchable, to make it correct, to make it authoritative, to get people out of their spreadsheets that you have stored away somewhere and unify that data so that it starts to tell a story. And we've been very successful in those efforts, but it's a massive undertaking because you have companies that have not necessarily thought from a data-first point of view and are now realizing the actual value of this data. So part of what we're doing is extracting that metadata, doing it in a way that's extremely accurate and authoritative, but also going as far upstream as possible to try to find, are there other people that are already collecting this metadata, and can we have them put into kind of a central database as opposed to everybody having their own little corner of data? Is there an effort to kind of reassess the value of the data where before just raw data in and of itself was liabilities, expensive to store, expensive to keep, and there was always trade-off decisions about what you keep, what you throw away. Now, there really is the opportunity to keep it all, and there's significant data outside, maybe beyond whatever the box office gate of the feature film with all the various distribution channels and ancillary things, and obviously, Disney is way ahead of the curve in terms of licensing and realizing value beyond just the core asset. But are there new ways now that those models are being worked in so that you can justify the additional expense of all this extra metadata and storage and infrastructure, which at the end of the day, you got to pay the bill to the data center. Absolutely, I think to the extent that we can use our data to tell our stories, to gain new insights, it is extremely valuable, and I think there are efforts around the company to not necessarily store as much data as possible, but to find what data is valuable and where it is, and we're finding more and more data that is valuable, because when you are able to unify it with other data, it starts to tell a story, and that's both data about our content, about our content performance, about our consumers, what type of stories we should and shouldn't be telling. So I think it's not just taking everything, but it's figuring out what data is actually valuable and then trying to derive as much insight as possible from that. All right, so 2017, what are your top priorities for this year? I can't believe we're a third of the way through 2017. It's just a little scary. It used to be like a new year question, I guess it's not a new year question than you are. I would say one of our main goals is really to advance our automation efforts. I think also to the extent possible to advance our metadata tagging efforts as much as possible, I'd say that's kind of top of mind at the moment. In addition to some other things, but that's some of the stuff we're thinking about. All right, great. Well, Avi, thanks for taking a few minutes and enjoy your first ever. Thank you. 2017. All right, Avi Swardlow from Disney. I'm Jeff Frick from theCUBE. You're watching us live from NAB 2017 at the Las Vegas Commissions Center. We'll be back after this short break. Thanks for watching.