New tech from a Toronto-based group is bringing life to the residents of Night City in Cyberpunk 2077
CD Projekt Red’s Cyberpunk 2077 is obviously a massive game. Players will be able to explore a vast city and its surrounding areas and fill a multitude of roles, but with that scale and setting comes people, people that don’t just talk, but do so in different languages. To keep players immersed and push themselves forward the developer has partnered with JALI Inc., a tech company based out of the University of Toronto, to create realistic lip-sync on a wide scale without the need for facial capture.
The drive for better facial animation and lip-sync techniques came from the already impressive job CD Projekt Red did on The Witcher 3: Wild Hunt. In a video demonstrating the new tech, Mateusz Popławski, the lead character technical director for Cyberpunk 2077, said that a large goal for this upcoming title was creating realistic facial animations without spending hundreds of hours actually animating. For Cyberpunk, the developer didn’t just want to improve animations past the quality of its previous title, it wants to bring those improvements to every single character in the game’s open-world – including NPCs that players will regularly walk past. That task is already monumental, but then CD Projekt Red threw in the additional challenge of animating faces and having lip-sync for 10 different languages, all with no facial motion capture.
This task brought the developer to JALI, and it ended up being a perfect collaboration. Specifically, JALI’s tech is a procedural, heavily customizable way to animate faces in a realistic manner. It takes into account all the things that we often overlook when we’re speaking to someone else – eye movement, brow movement, neck movement and the changes in our facial muscles when we speak. It also accounts for the different ways people produce phonetic sounds to create distinct profiles that match characters.
Here’s the progression of a face animated with all the tools JALI has at its disposal:
That slow evolution from completely robotic to somewhat human motion was created without any animators, exclusively through procedural animation. In some ways, it doesn’t stack up to the standard of extensive facial motion capture, like in The Last of Us 2, but achieving this level of detail without hundreds of hours spent animating is impressive.