In the fourth of our blogs with Gravity Media’s UK Head of RF and Speciality Cameras, Tony Valentino, we take a look at Gravity Media’s R&D programme and the exploration into using artificial intelligence in remote camera technology.
The rise of the robots
Last time we spoke, we finished off by taking a quick look at remote production, and that’s where we’ll pick things up today. Remote production – where you have just a small team onsite sending signals back to the production team at a remote hub – has been an increasing trend for a while now, and its popularity has certainly accelerated thanks to the pandemic. There are lots of benefits to working this way, not least because it minimises costs and carbon consumption, and simplifies the onsite rig.
But what role does our Specialty Cameras team have to play in all this? Well, using tennis as an example, most remote productions currently use a combination of traditional and robotic PTRZ cameras, with the latter being deployed on the outer courts and positioned strategically around the grounds, covering lower-profile matches and atmospheric shots of the crowds meandering around. Both those types of camera require an operator onsite – either physically sitting behind it, or, in the case of PTRZs, controlling it from a control room or the OB.
What we’re looking at now is whether these robotic cameras can be used for a complete sport production, remoting all the feeds back to the hub so that, potentially, there are no cameramen and women onsite at all. In this scenario, they’d sit at the central Production Centre with the director and the rest of the production team, which is far more efficient than travelling to the event location: no lost days sitting on planes – or in quarantine – and more availability to work on consecutive days. And as the kit is so small and light, it’s cheaper and easier to transport, too.
The art of the possible
There’s another layer that we’re exploring, which is to add artificial intelligence (AI) for those remote cameras. Looking at tennis again, this allows us to track players automatically ‒ following and recording their every movement ‒ and means we don’t need camera operators at all. Tennis lends itself well to this. It has a repetitive pattern: the players go to the baseline, serve, play out the point, then return to the baseline and do the same again. With that kind of scenario ‒ with two or four players in a small, controlled environment ‒ it’s easier to follow the gameplay; to design the algorithm to track those players accurately. This has the potential to allow coverage from far more courts at a tournament, and to reduce the amount of resources needed to acquire the content.
When you look at team sports like rugby and football, however, it’s a bit more complex. How do you track the gameplay there? Perhaps you’d focus on the referee, or pick three or four players at a time and track the ball in frame, but whatever you do, it’s certainly more complicated. With rugby, the ball often gets masked – in a ruck or maul, for example ‒ and the gameplay’s completely different; much, much harder to predict. So this is still very much emerging, experimental technology – imagining what might be possible, and then finding a way to make it work.
Putting it to the test
Speaking of which, part of our R&D programme here at Gravity Media has been to set up some remote testing with PTRZ cameras at recent live events, in a bid to ensure it is possible to cover some of these sports with these types of cameras effectively. For example, we were at the ATP Monte Carlo Masters in April, where a suite of remote cameras was controlled from our UK base in Watford. The test went really well, with the team able to fully control all the PTRZs remotely, although the delay still presents a bit of a challenge. Work continues to optimise paths to make remote operation truly viable.
There are other applications for this technology too, outside the main sporting events that we’re already used to watching. I don’t think we’re too far away from being able to use intelligent cameras to cover lower-tier sports without the need for camera crews onsite. So all of a sudden, non-league football and club-level tennis could find itself being made available to fans in a way it never has been before.
We’re talking about serving a different audience in a different way. What we’re exploring doesn’t compete with the Champions League final being broadcast on satellite TV; we’re taking about people on the move, using their mobile phones, tablets, etc, who just want to see their local team and people they know, maybe live, playing in the game. We think they’re not particularly fussed about slo-mo replays and the highest-quality images; they just want to their sport to be covered ‒ streamed, probably ‒ on one of the many platforms that are brilliant for this type of content. That would be amazing, wouldn’t it?
And it’s an expandable, commercial opportunity too, of course, because there are so many of those sports and activities, and there are so many potential viewers ‒ subscribers ‒ who would be eager to sign up. I’d be surprised if we didn’t see this theory becoming reality in the next couple of years. It’s a really exciting time to be in the industry.
It’s important to note in all this that we’re not suggesting that robotic cameras – with or without AI – are the answer to everything. These are additional uses; additional markets. There’s no possibility that the vital role of the camera operator can be replaced completely. The big matches and the big arenas will always call for the expertise and dexterity of the human touch, it’s just that robotics and AI do promise to open up some hugely exciting avenues for the fans. And if that’s the case, it certainly gets my vote.
For more information, or for support with your next production, email email@example.com.