The model takes into account the common tennis strategies and the tendencies of each player. From his training, for example, he knows that Djokovic likes to hit the ball towards the weak side of his opponent. Similarly, the model also takes into account how players tend to position themselves while waiting for their opponent to return the ball. This places Federer closer and closer to the baseline than Rafael Nadal, which shows how these two play the game in real life. According to the team that created the AI, it is this aspect of behavior that separates the project from previous attempts to create a system that can simulate the game of tennis.
With these features, the system can potentially create endless what-if scenarios. It can produce recordings of Federer playing against himself or Serena Williams. It is even possible to extrapolate how a match might have played out differently if a single shot had landed in a different location. In addition, the system gives you control over a player̵
Of course it’s not perfect. While the researchers did their best to hide potentially disturbing visual artifacts like changing lighting and players ‘clothing, there are moments when the clips look more like they were torn straight out of a’ 90s FMV game. As you can see from the clip above, fans and officials don’t move at all. Another factor that breaks the illusion is that the ball and players are not casting shadows. It also looks like they’re walking across the surface of the square. Still, it’s an impressive system that could have many fun uses.