Emma is Unity’s Lifelike ‘Virtual Human’ Who Can Potentially Mean Big Things For Games

The technology that goes into bringing video game characters to life may be getting a lot better thanks to a unique Ziva RT tool, which is being used to create “virtual humans” like Emma — a lifelike 3D character that can run in real-time in a game engine.

Ziva Dynamics, which was acquired by Unity earlier today, is developing the Ziva RT tool in the hopes of streamlining the process of realistic digital character creation for artists of all skill levels. The project is being spearheaded by Unity Director of Global Graphics Natalya Tatarchuk, who showed IGN a demonstration of Emma.

The character is already realistic, but rather than taking a long time to have each frame of animation processed by an incredibly powerful computer, Emma is able to run in real-time, with changes made to the animation reflected almost immediately.

This is significant because it can take a small team several months to create and animate a medium quality character for a video game. For the lifelike animations we see in films and games like The Last of Us franchise, studios use motion capture for the performances, but end up with animation clips that can’t react dynamically.

By comparison, Ziva RT has the potential to allow creators to take a 3D model, upload it for processing in the cloud, and get back a face puppet within an hour that can match the visual quality of motion-captured faces. Crucially, these characters can have their animations adjusted dynamically, making them ideally suited for games.

The goal of this technology is to make “realistic, lifelike character creation accessible and scalable for all artists regardless of skill level,” Tatarchuk says.

Unity’s animation tech takes another step forward

Another major accomplishment of this technology is its scalability and ability to run in real-time. These are biomechanical simulations with skin, muscle, fat layers, and fabrics involved, but Tatarchuk says the tech can theoretically support hundreds of characters at a time on, for example, a PS5. With level of detail scaling, far away characters can have their animations simplified to cut down on unnecessary processing. It can even scale down to run on phones.

Right now, Ziva RT is working for faces, but it will expand to bodies and creatures in the future. That said, it isn’t limited to human faces. It can be set up for realistic or stylized faces, like aliens and monsters — a limitation of Unreal’s similar MetaHuman tool.

Leave a Comment