Issue link: https://digital.copcomm.com/i/196542
©Atomic Fiction. Motion Capture ■ DIMENSIONAL IMAGING'S software tracks every vertex on the face mesh to deform the mesh directly. is important in the video game industry. In the interactive world, they have hours of animation. They aren't making a movie. And the video game character may be rendered in the game engine. " Although the company's animators and engineers often track markerless footage, they also work with data captured using markers. "For most game characters, you wouldn't need to track more than a few hundred markers, Edwards says. "Even though " your target mesh may have tens of thousands of vertices, you don't need to track thousands of points because in most cases on the game side, the character is not the actor. " Founded in 2009, Cubic Motion now has a core engineer- ing team of seven and approximately 20 technical animators. "Sometimes we might take part in a capture session, but we usually try to stay away from that, Edwards says. "And that's " the way we feel the industry ought to work. We don't want to get involved in capture, and we think maybe capture studios should not spend time analyzing if it's not their specialty. " Two groups within this visual effects studio concentrate on performance capture and facial-motion capture: the Virtual Production Group and a Virtual Human Group. Although the groups scale up with projects, each has a core team of 15 people, including the key artists who worked on the Oscarnominated Real Steel and Oscar-winning Benjamin Button. Gary Roberts heads the Virtual Production Group, which operates from Robert Zemeckis' former motion-capture stages in La Playa, California. "We service the internal Digital Domain teams working on features, commercials, and video games," he says. "And two years ago, we opened our doors to anyone who wants to use us for motion capture and virtual production. Most of our business comes from word of mouth. We've made massive improvements in our digital-character and digital-double pipeline, especially in facial animation. We're moving that development from features into games now." Creating an animated face typically begins with highresolution scans of an actor's face at USC/ICT. "We use these scans to help build our rigs," Roberts says. "We capture actors doing 160 to 200 poses with an array of 30 to 50 high-resolution video cameras. It takes about an hour. Then, we pick out the poses we need. We get an amazing amount of detail – good geometry, great texture and displacement maps, and diffuse and specular maps, as well. It's super important to scan and survey the actor at this stage because we know we'll get less-accurate information during the performance." Once the team has finished scanning the poses, they stabilize the digital model of the actor in 3D space and, from each frame of the performance, reconstruct – that is, "solve" – the facial expressions. The goal is to create a 3D model of the actor rigged with blendshapes that can reproduce the actor's facial expressions. The seated capture session at ICT provides information the riggers need to modify the sophisticated rig and match the actor's facial expressions based on the set of poses. "One of the most important things in getting a good character face animating based on performance capture is the actual face rig itself," Roberts says. "The rig has the biggest impact on quality. The actual capture side is important, but it's reasonably well understood. Transforming that information onto a character rig is the hard part. We've put a lot of work into understanding how a human face moves, breaking that movement down into isolated muscle movements, and understanding how they combine to create a facial expression. That includes, for us, lips, mouth, the interior of the mouth, the eyes, and the neck. In fact, the joke within our company is that we think the neck is the new face." So that the team can capture the movement of an actor's face and neck during filming, they have the actor wear a head rig – a helmet onto which they attach four cameras. "We're still big fans of capturing and reconstructing the 3D surface accurately, and for that we need more than two cameras," Roberts says. "The cameras are five to six inches from the face and outside the actor's peripheral vision as much as possible." That said, the studio is also looking at 3D sensors, such as the Kinect camera, and at single-camera, head-mounted rigs, like those from Mocap Design. Once captured, proprietary software reconstructs the facial expressions on the digital model for every frame of the performance. "You have to be careful how you do that," Roberts says. "The isolated movements need to be not only accurate, 14 ■ CGW Se pt em ber / O c t ober 2013