Hawkeye, digital twins, and player tracking

12:18>> Yeah. And so it it it turns out they hired this company called Hawkeye that's what what they do is Hawkeye creates digital twins. So on photo four you'll see the the players get scanned head to toe and you create a digital twin of their body scanned by a computer and then that digital twin of the athlete can be dropped into a virtual simulation of the game to determine their exact position based on 20 16 different high resolution cameras. So in photo five we've got the photos of the high resolution cameras and one of the one of the things that you learn in experimental science is like the error goes down by one over the square root of n. So in

12:59this case whatever if you had just one camera and you you were trying to track the position with 16 your error goes down by a fourth. Okay? Four times smaller. >> Yep. Yep. [clears throat] >> Okay, that's quite good. But that's still not enough. The real magic here is the digital twins because it's something that we do a lot in science which is we've got a model of our system and then we've got data. We fit the model to the data to get really high precision. We do this a lot with atmospheric um transits of like exoplanets, right? Like the exoplanet goes through, we get some spectra. The spectra is noisy, but we try to fit models to that spectra to really understand what is going on. In this case, our model is the player's 3D

13:41digital twin, and we're fitting that model to the data from the 16 different sensors. >> I I think most people didn't even know that they got scanned into 3D for reasons other than to be put into the FIFA Excuse me, uh FC 26 the game because now FIFA no longer gives EA Sports the rights. So, it's no longer FIFA insert year, it's now FC. But like for most people when they think about like they think about motion capture and 3D scanning, that's the context. >> Yeah. >> I think it's a very important detail that that's why when you see those little die like the little example photo we just showed, >> Yeah, yeah. >> like it's used >> the 3D model of the player. >> the player itself. >> Of the player itself. They're using like computer vision to actually like fit the 3D model to the player. And this is

14:23something that like we do all the time like even in neuroscience and behavior experiments, like when you have like a rat running around in a maze, you've got a camera that tracks the rat, right? And then there's you can create a 3D model of like the rat's joints to figure out his body position, correlate that to the neurons that are firing, and so on and so forth. It's used a lot in behavior experiments. I'm sure like similar technology is being used to recreate the player positioning at every given moment. So, >> Cuz cuz uh Ivan Perišić and Erling Haaland are very different players. >> Yes, exactly. >> You know. [laughter] And so, it matters. >> Yeah, and it is kind of cool that like they have that kind of granularity. So, that's the where >> Mhm. >> of where the players are. Now, what about the when? Right? Because I said

15:03the 30 frames per second is not going to do you any good. Even the 50 frames per second if you have a high-speed camera is not going to do you any good.