FIFA 22 brings machine learning to the field with HyperMotion

As Sam Rivera explained to me, the success of FIFA 22The new animation technology will be seen in what was not recorded during a groundbreaking motion capture session, which involved 22 players, all playing a soccer game from start to finish, earlier this year.

“We started working on an algorithm about three years ago,” Rivera explained, FIFA 22Leading Game Producer at EA Vancouver. “What that algorithm is doing is learning from all the data for that motion capture shot: how the players approach the ball, how many steps they take to get to the ball, it’s three long steps and one short step; What is the right angle, with the right cadence, to hit that ball correctly? “

Then, Rivera says, “create that solution, create the animation in real time. That is very, very cutting edge technology. This is basically the beginning of machine learning that takes over animation. “

In a dozen years covering sports video games, I’ve found their almost annual call for new animations – often a record number, more than ever – to be a marketing cliché. Recording a bunch of fancy stunts doesn’t necessarily mean that regular players will see them. Almost a decade ago, NBA Live designers told me sifted the remains of the abandoned NBA Elite eleven and discovered tons of amazing blocks and dunks that the game engine simply would never deliver. The lesson: taping these things is fine, delivering them is where the rubber meets the road.

On FIFA 22In this case, your motion capture team could have captured the true anguish of a goalkeeper hit by the decisive scoreboard, in a 4-2 match between two real rivals from Spain’s RFEF First Division (the third tier of the nation). But neither that, nor the desperate play of CD Gerena that led to Atlético Sanluqueño’s crushing counterattack, should be the reason why players at home feel they are seeing a more realistic presentation of football, or that they find it more fluid. and responsive while playing it.

HyperMotion, as it’s called, because, yes, marketers will be marketing these animations as well, it allows Rivera and his development colleagues a proposition both ways when it comes to movement and player interaction, he said. “In the past, we used to prioritize short animations, so the game is responsive,” he explained. “If you’re in a long animation, the game looks good, right? But if the situation changes [in the middle of the animation], a defender is coming, you’re stuck, they’ll probably tackle you, and that doesn’t feel right.

“With access to HyperMotion, we can put in longer animations, longer ball control animations,” Rivera said, “but the technology allows us to transition in the middle of the animation to a different type of animation, if the situation changes. “

More specifically, HyperMotion will look at things like the defensive formation that maintains its shape and moves in concert more consistently. That will probably be the first and most visible piece of evidence players will see from the recording session of the 22 in the new game. In previous editions of FIFA, the AI ​​would normally direct players in groups of two, maybe three, depending on who was closest or if someone was manually called in for support.

Players of FIFA 22 You should find the defense more difficult to separate or challenge one on one. And in response, off-the-ball players on offense will move more naturally to support a run, perhaps ameliorating a common frustration felt on higher difficulties or in online multiplayer.

“Now that we’ve seen some comments about FIFA 22 and the closed beta, people are definitely saying that there are a lot of features that make the game different, it feels different, “said Rivera. “We were recording something with our professional [esports] players, and they were trying to play a game similar to FIFA 21. And they didn’t necessarily score many goals. Even in the opportunity creation game, you have to think a little differently. All the things that the animations changed, that we changed in the game, in terms of positioning, very different experience. Cast [experience] it was, ‘It feels fresh.’

As Rivera put it, this is “the beginning of machine learning taking over animation,” and the EA Vancouver FIFA team is aware that their work could have applications in other EA Sports products, if not others. Electronic Arts games in general. However, in the first year of HyperMotion implementation, Rivera said his team is fully focused on getting this right within FIFA 22and less about using that as a proof of concept.

Most of Rivera’s work implementing HyperMotion was not in the algorithm (which was done by an advanced work division at EA Vancouver) or in motion capture. Rivera and his team were working on building conduits and processes to accept, interpret, and use whatever his machine learned and spit out. Those processes are specific to his game, he said. If Madden NFL designers tried the same in their sport, they would have to develop their own. And Rivera’s first client is the FIFA 22 player, anyway, not another developer in your company.

“At a high level, what we hear the most is that people wanted more differentiation” between the players on the virtual field. Each year has certain subtleties, problems and feats, he said, but “we knew that what people wanted was a leap in terms of differentiation and realism, to make sure the game looked better. We knew then that we needed a bigger investment in terms of animation and technology. “

That took three years, capped off by 90 minutes on a field in Spain. “If you multiply that 22 times, that’s what allows us to bring more than 4,000 animations to the game in just FIFA 22? “

Ah, there it is! The count of animations. But Rivera is legitimately proud of it, anyway. “It’s a record for us,” he says.

List file is Polygon’s news and opinion column on the intersection of sports and video games.