This neural network could make animations in games a little less awkward



The graphical fidelity of games these days is truly astounding, but one thing their creators struggle to portray is the variety and fluidity of human motion. An animation system powered by a neural network drawing from real motion-captured data may help make our avatars walk, run and jump a little more naturally.

Of course, if you’ve played any modern game, you may find that many do already — but this is painstaking work accomplished by animators working from libraries of motions, linking together all kinds of contingencies — what if she raises her bow as she’s going up stairs and crouching? What if she gets hit while trying to balance on a narrow beam? The possibilities are quite endless.

Researchers from the University of Edinburgh and Method Studios put together a machine learning system that feeds on motion capture clips showing various kinds of movement. Then, when given an input such as a user saying “go this way” and taking into account the terrain, it outputs the animation that best fits both — for example, going from a jog to hopping over a small obstacle.

No custom animation has to be made transitioning from a jog to a hop; the algorithm determines that, producing smooth movements and no jarring switches from one animation type to another. Although plenty of game engines do improvise a little for things like foot placement and blend animations, this is a new approach that could prove more robust.

Machine learning has been brought to this space before, but as the team shows in the video, the systems produced were pretty rudimentary, showing the wrong type of movement or skipping animations because they weren’t sure which to use — or deciding too strongly for an animation state and producing jittery motion.

Subscribe to our Channel