The VideoVerse
The VideoVerse
TVV Ep 13 - Scaling sports video, and statistical data in the cloud.
In this episode of the Video Verse, Casey Bateman shares his insights on the challenges of migrating from H264 to HEVC and optimizing live streaming experiences. He also talks about their recent acquisition of BlueFrame Technology and how it will help teams make money by streaming out games to fans. Tune in for an interesting conversation about the future of video!
Watch the full video version.
Learn more about Visionular and get more information on AV1.
[Announcer] Welcome to The VideoVerse.
Zoe: Okay. Hi everyone. Welcome to the The VideoVerse Podcast. So here's Zoe. I'm going to become the host of this episode, and I'm having here, David, my colleague, as a co-host for this episode. And then I'm very honored, actually, very excited to introduce a guest to this episode, Casey. I won't steal the thunder 'cause I will like Casey to introduce himself, as well as the company, "Hudl," behind that he has, I think he has been working there for many years, even though I have the number. I would like him to speak out, and then we'll again talk about the video technologies. So Casey, happy new year, and thank you for coming down here. So you can just have an intro of yourself as well as the company you're working with, whatever that you want to put up, okay?
Casey: I'm Casey Bateman. I've been working with Hudl here for actually about 10 years. It'll be 10 here in April. I am a Principal Engineer in the office of the CTO at Hudl. And a little bit about Hudl... We are a sports technology company. We focus on all different areas of sports. We have over 40 different sports that we serve. We currently have over, it's about 200 million, or it's actually 200,000 teams that we're serving and about 6 million users that we're serving worldwide.
Now we focus on multiple different levels of sports, all the way from your professional teams to your Division 1 sports, down to high schools, and then down to club and youth sports as well. So we kind of cover the broad spectrum of levels of sports there. And I would say the majority of our business is actually focusing though on the long tail, the club and youth and the competitive market. We're currently sitting at about 98 to 99% penetration in the American football market in the US a year. So that's kind of our bread and butter sport and where we got started too.
David: Wow. I think talking a little bit about the background about how Hudl started, it'd be great if you could share a little bit of the origin story, and more about what that sports video use case is. It's not just reminder, I won't speak to it, but it's not just live streaming where there's lots of other cool things you guys are doing with it. So maybe we could talk to that a little bit. It'd be great.
[00:02:40 Cool things about Hudl]
Casey: Yeah, absolutely. So Hudl started back in actually 2006 by three guys. It's David Graff, our CEO, John Wirtz, our CPO, and Brian Kaiser, our CTO. They were students at the University of Nebraska and the Raikes program back there and actually, they were roommates. John Wirtz and Brian Kaiser were roommates, and then Dave Graff and our current COO, Matt Mueller, were living across the hall from them.
Now Dave Graff, back in 2006, he was working with the American football team, in the athletics department with media relations. And he had good access to the head coach at that time at University of Nebraska, which was Bill Callahan. Now Bill Callahan came from the NFL. He was really interested in ways to incorporate technology into his team. And if you knew anything about the football landscape and how they were viewing video at that point in time, you know that technology wasn't like where it is today.
All video back then was... Video review was done on DVDs. They would have countless people who would be burning DVDs of film and sending them out. And you would have players that would take home 10 or more DVDs that they'd be required to watch before the next day, right? And this wasn't just a college level problem. We would have high school coaches that whenever they'd wanna watch scout film, they would have to burn a DVD and go drive to the next town, or meet halfway just to drop a DVD off. And so they could actually review their video.
As great as DVDs are and stuff, as you may know when you're trying to go watch a DVD, you're skipping back, skipping forward. You don't get the benefit of watching things reverse and slow mo. You sometimes overshoot where you want to go. It just kinda jumps, right? So it just, it wasn't a great experience for watching video. So, rewind back here, 2006... The founders had an idea to try and bring that video analysis to the cloud and stream video down. And so they created the first iteration of Hudl. And Hudl was like a video review tool that was first used at Nebraska.
So in 2006, Nebraska became our first team. Now, it was a little slow going after that. We actually had a, we had another team the following year. So the following year, actually, the Huskers had a bad year and the coach, Bill Callahan, ended up leaving and he ended up going to the NFL. The good thing was he brought Hudl with him. So our second customer was actually in the NFL to the New York Jets. At that point in time, Hudl was very much focused on that elite space, right?
So... After a while, we didn't think that was gonna work, just because we had a lot of high school coaches reaching out to us asking, "Hey, can we use this? This sounds great. This is a great tool." At that point in time it was really expensive. We didn't have AWS. At that point in time required a lot of onsite servers and management. So it was about a $50,000+ operation just to get Hudl installed, which is just a massive barrier to entry for all these smaller schools.
And it wasn't until about 2008 that Brian Kaiser decided, "Hey! We should pivot a little bit and take a look at this really cool technology as Amazon's web services. And try and put everything, build Hudl.com, put it on the cloud and have streaming there." And that's, from there, we were able to launch a beta and immediately got 10 schools on board and just saw a rapid growth to where in 2009 we jumped up and had 2% of the market share.
Then jumped up to about 2000 teams in 2010, over 5,000 teams in 2011. And then it just kept growing and growing from there, until eventually, we just captured the football market and stuff, which is where Hudl kind of originated in American football. And that's a little bit about where Hudl came from and kind of why we were serving the sports, and why we decided to tackle that front.
Zoe: I think this is very amazing in the part that the Hudl story actually reflects the whole internet video streaming history. Because nowadays we take everything for granted. We think like AWS has been here forever, cloud solutions have been there forever. And I don't remember what is the last DVD, the physical DVD, that I watch movie, we watch a drama from. 'Cause I only watch videos right now on my laptop or the phones. Assume that they has been there forever, since at least a long time ago.
So even though I think you mentioned that Hudl was, it's about 2006, right? So right now it's about, this is going to be 17 years, I guess, but we don't feel that. We thought, "It has to be there." So your case is just reflect the whole history to me. I was just wondering because you mentioned that, not only like sports, you start from the football and then right now you already spread that to more than 40 different kinds of sports.
And so regarding, let's talk about the video content 'cause for any video technologies, they are actually can never be general. You related many technologies are very specific for different contents. So we like to hear what you feel when you work on this video 'cause you have been working on this for 10 years. Sports, how special it is? And then, what is really the customer when they view, right? Because you provided the review tools for them. So what do they really expect? What kind of sports company, quality, resolution, all kinds of characteristics that you need to provide? And how special this kind of sport content as opposed to other contents that we watched?
[00:09:29 What is user's expectation]
Casey: Mm-hmm . Yeah, absolutely. So our customers, they're the ones actually uploading video to us. So you can kind of think of us like a YouTube for their video content, right? So they're uploading content. They expect that content to be available as soon as possible. They expect it to be uploaded as soon as possible, and they want as much hands off as they can on the content.
Now as for the different customers we serve, we actually serve a wide variety of users, right? We have coaches that like to view our content from the program review, so kind of like from a review perspective. They like to look in, check out formations, see which players are doing poorly. They like to mark on the video. They like to kind of stream it down. And it's mainly for team review purposes at that point in time, and pair that up with some of their other analytics and statistics.
Now we also have the athletes. The athletes, they want to, they view it in a different way, right? Athletes like the view that content and more curated to them, as their individual plays in what they did poorly, what they did great, how can they learn from that. But then, that's kind of from the team view. We also have kind of a fan view. Athletes like to use the software to tape their great plays, and they like to make custom content that they can share out with their fans, really kind of build their brand, build their recruiting profile so that they can get some great, they can get some great recruiting exposure out in the field. And we've actually had some great, we've had some great stories about athletes who, because they used Hudl, ended up getting themselves a D1 or D2 scholarship to go play in college, which was really cool to hear.
And then, we have fans, right? Fans who like to watch that, they like to see their favorite athletes follow them. They like to take a look at that video and favorite it or share it with people. So we've kind of got a wide variety of different users and they all will actually digest the video differently. Because they digest the video differently, we use different technologies to serve each individual use case, right? And each different sport has actually different requirements as well. So jumping into that a little bit, we've got football, right? We have American football, it's a very stop and go. I mean you've gotta play, it's gonna last about anywhere from 5 seconds up to 40 seconds, right?
That doesn't really bode very well for a HLS stream, right? An HLS stream isn't gonna work super well in that situation because we've got a very short form video that coaches are digesting there, right? They want to... They use it a little differently with reverse playback and fast forward playback. It works a little better in those situations to do traditional streaming. But then you have sports like soccer or European football, or basketball, which are much more flow-based. Plays don't just stop and go, right? In a situation like that in basketball, for instance, you can have a steal, a make, a free throw and everything, and the the play just, it just continues and continues and continues. There's no real stoppage of play unless there's a timeout, right?
So there's instances where you can have 10, 15 minutes of just continued video. When you're talking about coaches trying to watch 10 to 15-minute long MP4s, we all know seeking through 10 to 15 minutes of video is a problem, right? If you need to seek to a moment halfway into there, I can't sit there and wait for half the video to download just to be able to watch a certain point in the video. So we use HLS content for long form video 'cause it really speaks to just the ability to quickly jump to different points, to be able to adapt to their various bandwidth needs, and just works really well for more of those flow-based sports.
Zoe: Okay, so you basically mentioned that the HLS streaming is quite well fit into the flow-based sports kind of content because there's a need for like random access, for example, and that be quite friendly. And you also mentioned that you have actually a varied variety of view base. You have the coach view, and then you have athlete view and fans view. And so meaning that you must provide a different, different tools, of course, based on your platforms.
And behind that, I believe for example, flow-based, sometimes they go with, like you mentioned, the 15 minutes of a video, right? And for fan base, they really want to see something like exciting, still or some, a good moment. So believe that as some technologies, you must have leverage. Like say, people say, "Okay, how do you leverage AI technology to dig out those very exact moment automatically?" Because based on you just described, you have covered such a variety of sports and then from, you also mentioned from the colleague, from high school, and the colleague, to even for the professional. And then that means that you actually have a big athlete-base, and then they also must spread the words, at least with the motor mouth you have expanded to a larger variety of users. So that means you must handle a large volume of videos and with each video have different characteristics. So there got to be some AIs behind that?
[00:15:28 How to meet the needs of a large variety users]
Casey: Yep. Yes, absolutely. We do... So when you're talking about those exciting moments, right, that fans really want to capture, we really source determining those exciting moments from a couple different avenues, right? First and foremost is the fan, right? That's how we can scale that, is fans can build their own brand, fans like to mark. We give them, to provide them tools, whether that's on their iOS device, on an Android device or on through our web portal to create a highlight, throw on spot shadows, throw on special effects in music and really make like a really unique reel of highlights that just showcases their talent. We've also got, we've also got some more smart features, right? So we've got some... We do have some AI, one in particular that I'm thinking, that actually listens to crowd noise. It actually determines based on crowd noise,
And how explosive the crowd gets on how exciting a different point would be. Now that is actually detected some pretty cool plays, like in buzzer beaters and stuff in basketball. And we will mark those as highlightable moments. And what we do, and one of the cool things about Hudl is it's not just video, right? One of our big powerful part about Hudl is we also get a lot of data, right? And that data comes from coaches, it comes from our Assist product, where we break down individual plays and moments in video. And we can take all of that data, bring that in, pair it with information we get from our different AI systems, and we can determine really cool, automatic highlights out of that.
So we actually have a tool that we rolled out, I wanna say back in 2015, that would look at game data, look at those moments and put together a highlight reel, like a custom recap of a football game. And we expanded that into other sports as well. So we've got all of these auto-generated pieces of highlight content that we can throw out just based on the data that we're getting in. And we can use that AI to determine the exciting moments and put special effects throughout the video to make it really feel like these users are being starred on ESPN.
Zoe: So basically, with these tools, you make something very professional, but out of the tools you don't have to rely on very professional hands. And I also heard, because we talk about videos, but definite video is not independent word format, right? So you mentioned that it's being facilitated by the audio signal like the cloud noise, and also a lot of metadata associated with the video.
And actually, that's also very, very essential 'cause we're talking about the all kinds of different format data that come together along the timeline. And then when we talk about video, it's not only the same thing is, there's a lot of associated signals along with the video that we have to consider. And this is, we see a real usage down there, outside your use cases. I just want to, because knowing you Casey, I want to back to a little bit of personal, 'cause you joined Hudl for almost 10 years. You mentioned that by April, this is mark your 10th anniversary. And then this company is still young, right?
You mentioned that it's founded 2006, but then only until 2008, and they found that there is a great potential, not just one university that the founders graduated from, and/or one NFL users that the coach brought to, and they start to scale. So that's already bringing the company to the year 2008. And you actually join, I believe, based on what our previously discussed, on the early stage. And we are just wondering how you came into Hudl, how you have been working in this, I believe, exacting, exciting video field. Talk about little story about yourself.
[00:20:01 Story about Casey]
Casey: Oh boy. Well, so I guess going way back, I wasn't really into computers to begin with, or really into programming. I didn't have a... I was one of your classic, single finger like typers there when I was getting into college and...
Zoe: But no worries, I have to mention that actually I have a, I don't want to mention them, but I, I do see some very senior professors still use the single fingers at time.
Okay, go ahead.
Casey: Hey, for what it's worth, I was really quick at that single finger typer so... But back then, I wasn't even really, really all that interested in computers. I was really interested in the medical field. I kind of wanted to get into the medical side of things. And it wasn't until my mother, who was a cobalt programmer for a company called, "Information Technology," in Lincoln, Nebraska, it was a big banking software company, it sounded like a really cool place to work, kind of mentioned, "Hey, you ought to give programming a try. You gotta like see what you think about it and and stuff." And that really got me down this path of really loving technology. I joined a college out here, where they had a really hyper accelerated, almost code-camp-style programming school. And got myself my first job in sitting out in tech building banking software, which doesn't sound all that fun, banking software versus where I'm at now with video and sports.
Yeah. So fast forward to how I got into Hudl. I worked at that company there for six years, and a good friend of mine I kind of mentioned, "Hey, there's this up and coming company in town called, 'Hudl.' I think you ought to take a look at it." It kind of looks cool. I gave it a shot and the rest is history. I ended up falling in love with the company. When I first came to Hudl, I actually knew just next to nothing around video technologies, next to nothing on video streaming or anything like that.
And there was a hole that needed to be filled in the company of someone with that kind of video knowledge, who was really interested. And I decided to give it a shot. Ended up falling in love with it, where our first project was making an auto-generated piece of content for coaches trying to raise money for their schools, where we, essentially, put together a bunch of highlightable moments from video, put some sad music and real tearjerker stuff on it and made people want to throw money at the screen, sort of thing to just donate to schools. And I just, I fell in love with the video technology at that point in time, and that was kind of how I got to Hudl and got into the position I was at.
David: That's pretty cool. Go ahead, go ahead.
Zoe: I was just saying that ever since then, you grow a lot and then you are hooked in this field.
David: Yeah and I was gonna say, not only has, you've grown obviously, but I'm, the volume of video has grown tremendously over those 10 years as well. So I'm curious, Casey, we talked a little bit more about A, dealing with that volume of video, as well as, you know, how you think about the compression of that video, right, the codecs that have come and gone as well.
Casey: Yeah, absolutely. So obviously, as we scaled, we needed to make sure that we were, compressing our content a lot more, making it easier for coaches to watch things, as well as kind of managing our cost side of things. I mentioned, we focused on that long tail of sports. So there's an entire area of our strategic planning around this that just goes into storage of just that content and how long we store the content, how efficiently we want to compress our content to make sure we're managing costs on our side and stuff.
We currently have about, I believe it is 73 petabytes of video content that we have stored out on Amazon S3. So it's a really large library of information. And when we first started out, we kind of always stuck with H.264. So H.264 is a video. Kodak has been at the core of our business since the beginning. And our audio and stuff, we were originally in WMV containers, so you gotta love WMV, as a good old fashioned package there. But it worked really well with our solution, which was heavily based on Silverlight at that point in time.
So long before the HTML5 video even came out and stuff, we were using Silverlight and WMVs. And so, since then we've obviously grown and changed up mostly in the technology space, right? We moved to HDML5 MP4 videos, AAC audio, to now we're kind of moving into a more modern phase with those flow-based sports, we jumped into there. We started getting into HLS, started as MPEG-TS over HLS video. And since then we're exploring getting into fragmented MP4, now over HLS and hopefully getting into some DASH in the future, if that's something that makes sense for our business, obviously. And hopefully, getting into some newer codecs too. Obviously, because we've got a lot of MPEG-TS right now, we're kind of stuck on H264.
Once we get into fragmented MP4, we've got plans to look into AV1 and other codecs as well to just kind of help compress that video even further, help manage our users' bandwidth concerns there by making sure that video is really easy to stream, which is a really important part. Some of these coaches, you know, not every coach is going to exist in a big city with great internet. We do have coaches from small towns, where their only form of internet is gonna be line of sight. And so they, you know, they may get 12 megabits per second, which we know is not constant. It'll drop to sometimes below a single megabit. And we have to make sure that we have technologies in place that are gonna help them to still maintain a solid stream at those pretty variable bit rates and bandwidths.
David: Yeah, very cool.
Zoe: Yeah, so basically the great volume need to be precise and then because you again mentioned, I think maybe already for the second or third time, that you are handling the big volume of long tail videos. And also you mentioned that an users' manual conditions could vary. So that's a typical, we are doing compressions, that's a typical compressions.
Actually, what you said also send a strong signal then to whoever working in this field, right? So meaning that, you're looking for good technologies, you can further compress the video, and then squeeze out more bits, but at the same time, because you mentioned you have a huge volume of data, just like a sports YouTube platform, and that means that you need to process something faster and with higher density, and then the latency actually, because you also mentioned that the users want to watch the video right away. So you got to process fast and the turn around time got to be short enough. And those are the typical issues that people talking about, but here reinforce about very detailed use case.
And so, we like to actually, specifically, talking about the accountant, like back to sports cause you mentioned this is sports. So sports, meaning that there's a lot of motions, and then I think, for example, because you have fans, and the fans sometimes they really obsess with certain players. And when they play, I do like to watch American football. I still such a soccer, 'cause we just finish the world cup, and then people talk about the soccer or the European football a lot. And sometimes we, of course, like messy everybody chase after that, but sometimes we chase after like the Jersey numbers. I think you also mentioned that at one time again. So this is some very specific content inside the sports video that may need to be precise, or at least paid attention, specifically.
So we like to hear something that, based on all the things you have worked out, why sports videos are so much different from... For example, right now we are doing the conferencing videos, and we watch entertainment videos. We like to hear something like that.
[00:29:26 Sports videos are different]
Casey: Yeah, absolutely. Maybe... So yeah, I mean you kind of hit the nail on the head when you were talking about that earlier. It is different from sport to sport. There is a lot of varying factors in motion that kind of play into each different sport. What we care about most is definitely Jersey numbers. What people care about the most is Jersey numbers. And the reason why they care about that so much when we're getting into video analysis, is that it's the identifying piece of data that you can associate with a player, right?
When you're watching a video, when you're watching that from a coach's perspective, an athlete's perspective, the first thing that's gonna jump out to you is gonna be that number. It's likely the largest piece of identifying content that we have on a player, right? And we have our analysts on our side through our Assist product, who will watch that video and use that jersey number 'cause they don't know much about other players on the team. They don't have that context of knowing these people for such a long time.
They use that to identify who did this, like who scored the goal? Who made the pass? Who had the penalty? All of these different things. That jersey number is just so important to what coaches and what athletes and what our analysts want to see. Now... Yeah, you kind of mentioned, as I mentioned earlier, you mentioned a little bit about... Obviously, motion plays a big factor, but also lighting plays a big factor on how much, on how much of the bits need to be allocated in the stream to get a clear jersey number, right?
Motion obviously takes a whole bunch out of an encoder, just to capture from frame to frame. There's a lot of changes, less reuse. When you talk about lighting, that can affect too, your ability to clearly see those. So even if you've got the clearest jersey number in the worst light, the slight variation in bit rate and blocking in the frame is going to make it harder to recognize, versus if you had really good lighting. So you know, we've got that. We've also got distance in the field plays a big factor in too. So I mean in soccer, specifically, right? You've got, usually the feed is coming from the press box or up the top of the stadium, from a camera or static camera there for analysis. So all of your users are very, all of the players are very small on the field, right? When you've got that smaller, smaller players, it's very fine details, right? And so being able to focus in on that, it just becomes harder.
So we tend to have to provide tools for our analysts to kind of zoom in a little bit, and if there's any blocking or any, any poor quality and poor defined lines on the foot, I mean when you blow up an already small image, it's gonna be really difficult to see there. Then you have, you know, a whole, there are just so many factors... We have a factor that we like to call, "The shaky cameraman factor," and somebody who's been, who's possibly on their 12 cup of coffee or no sleep, and they're holding the camera. They're just shaking violently and going left and right and stuff. That just makes a huge effect on the video of whether or not we can get clear, clear lines around players, clear jersey numbers. It's, yeah, it's, it's a pretty... It's a problematic area for compression in us. There's not generally a one-size-fits-all approach that is always gonna work in those situations.
Zoe: I think you already attached, actually, the mostly essential challenges to video compression or video processing 'cause you mentioned the character part. There's camera motions that is sometimes have to associate with the video. You also mentioned the video content by itself, and you also mentioned the lightings. Most of the time the lightings are also very challenging because those two video compression are motions as well, even though to humanize, they'll think that different, in the different perspectives.
And back to, because you mentioned camera... So we also noticed that how do not only provide cloud solutions, right? You also, I see in the website, you also have provided the camera, the hardware solutions. So we just like to touch a little bit before we wrap up is how you think the hardware, for example, the capture side, together with the cloud, software-based, algorithm-based solutions, how that come to whole package of solutions? Like how that, how meaningful that is to Hudl, or to add to Hudls' end users?
[00:34:33 Hardware solutions]
Casey: Absolutely. So the hardware side is actually one of the more exciting pieces. I actually believe, and I believe the company also believes, that the Focus camera is kind of the future of capture and upload of sports, to the Hudl platform. There's just so much more that we can do with these dedicated hardware devices that we built in-house. Coaches, from the coaching perspective, it's really a set-it-and-forget-it sort of product, right? They're able to schedule recording. We're able to use great technologies, like Amazon's internet of things, to be able to schedule and push down commands, update live with new software updates, all of these cameras. And from a coach's perspective, they're done with the game, they get on the bus ride home and their video is already live and ready on Hudl.
And in a lot of situations, by the time they're home, they may even have data from our analysts already on their video. So they get back home from the bus and they've got their video with data already, like already filled out. Then these cameras, they're just really cool pieces of content. They've got some pretty powerful hardware on there, where we do, we do some hardware encoding. We also do some computer vision directly on these devices. The devices kind of capture a panoramic view of the field, right?
Whether or not it's indoor, or a panoramic view of the basketball court or a panoramic view of a football, softball, soccer field. And using computer vision in some in-house algorithms that we've generated, we're able to track the motion of the game, track the ball carrier, the ball, and create a tactical feed of content, right? So we're able to actually get a more tactical view that's more zoomed in and upload that. So I mean you've really got this, essentially, the perfect cameraman device out there that really controls motion, really ensures that we've got a steady image out there.
And it's a game changer when we're talking about it from the compression side and in sense that it farms out our encoding costs a little bit, right? We've got these devices that get to perform the bulk of the initial encoding for our content, but it's also a game changer in that we set a standard for the exposure, the lighting and everything else that really plays into some of those very, variable factors of video coming into Hudl. We can kind of control that now. It's just, I believe that the Focus camera is going to be a very large part of our business moving forward.
David: I'm curious... Now as we head into 2023, are there any other specific trends or maybe challenges you see coming up this year?
[00:37:41 Specific Trends coming up]
Zoe: Yeah, that's also something that I like to ask before we wrap up this, 'cause this is just happen to be the beginning of the year. So we're looking forward to your views and from your perspectives.
Casey: Yeah, absolutely. I know one of the big challenges this coming year is gonna be, we're diving deeper into the HEVC space, the H.265. That's gonna be a challenge, kind of our first step out of the old H264 world and kind of moving into a more modernized HLS streaming offering there, when we're moving into the fragment in it before. It's gonna be a challenge for us as we kind of migrate away our old understanding of video and how we're gonna be processing this video to these new packaged formats, these new standards.
So I think that's gonna be probably the largest challenge. And then I would say, probably, another big challenge for us is going to be, we're getting deeper into the true live streaming experience. We acquired a company recently called, "BlueFrame Technology," that kind of focuses on live streaming for high school sports and clubs. And we're really trying to explore that offering this year as a way for Hudl to provide a revenue share model that makes it profitable for our teams to stream out their games to fans, right? And kind of make Hudl pay for itself, not a fan by allowing them to make money just by playing their games.
And so I think, I think the live streaming and learning how we can optimize that experience for fans and kind of optimize that entire viewer experience, when you're looking at it from a live versus VOD perspective, I would, yeah... So I would say, from the live streaming perspective, the big challenge there is going to be that we need to really optimize for live playback versus VOD playback. And that's, you know, we're shifting our mindset moving from VOD to live. It's a different world when you're talking about caching, you're talking about content delivery of live content versus content delivery of VOD content and stuff. You don't get a lot of those nice features of pre-caching way ahead of time in live like you do with VOD. So you gotta, you gotta rethink some of your approaches to making sure you get that fast join time, that low latency delivery. So I think it'll be an interesting challenge for our team to take on here over the next year.
Zoe: Thank you for actually taking us, looking forward to the year of 2023. And then we really hope that our role, and we believe that video will be further growing. And then we just really, just best wishes with actually you, Casey. And to celebrate your anniversary, and now also wish the best with Hudl. And thank you for taking your time with us and to share the stories behind very special. Actually, in some ways, it's a manifest a lot of things in common, but then now we actually hear a detailed story, and then specific experience from your side and we really appreciate that. Thank you for having the time with us.
David: Thank you Casey.
Casey: Thank you for having me.
Zoe: Okay, that's, thanks for everyone.
