Issue link: https://digital.copcomm.com/i/1505114
Jeremie Talbot and Junyi Ling helmed the film's character supervi- sion team, each handling different phases of the process. "I'm on the modeling and rigging side," Talbot comments. "I'm at the front preparing these puppets for animators to perform with. Animation does their work, and then Junyi's team takes the output and actually makes it look photoreal. So we were kind of a character sandwich." The project's pipeline followed an innovative path that encour- aged collaboration between teams. "There are a lot of dynamics that actually flow all the way from animation to rigging to effects, and then all the way to lighting," Ling explains. "Everything's constantly moving." This workflow allowed the teams to solve problems in a highly collaborative way. "Every department had to interpret that in- formation through the lens of their department so that it was clear," Talbot adds. For example, if the lighting team received a version of Wade that didn't quite look right due to his transparency, they would work on a solution. Pixar's research team has also been developing a "curvenet" tech- nology, which was a major component of Talbot's workflow. "It's es- sentially wrapping the character with a bunch of curves, and you can use those curves to deform the character," Talbot notes. From the very beginning of production, Ling and Talbot knew that the char- acters needed to be very flexible. "We wanted animators to be able to control any part of the character, not just where we decided the joints would be, but all over the place," he continues. "They helped us add an interface to that curvenet that was really animator friend- ly, so they could grab any points in these curves that were wrapped around the character, manipulate them, and time them to create these different silhouettes and different profiles." This technology was very useful for adding nuance to Wade's movements. "When he moves really quickly, they kind of smeared that shape using the curvenet controls that we gave them to give him a little more of a watery trailing or leading edge," Talbot explains. This also allowed animators to quickly test concepts with fiery and watery elements. "We wrapped little balls of water and little licks of fire with those same curvenets so that animators could grab them, shape them, and manipulate them in any way they wanted to help them get their ideas out there really quickly, even if they were rough," he adds. Talbot and Ling also harnessed the power of machine learning for their character supervision workflow. The technology is based on a technical paper that explained the concept of stylizing volumes with velocity and density fields. "If you give it a painting, it will change the flow and the density to match that painting as closely as possible," Ling explains. "Essentially it turns your idea into something that is ac- tually flowing and moving." Directing Animators Allison Rutland & Gwendolyn Enderoglu Allison Rutland and Gwendolyn Enderoglu, along with Benjamin Po An Su, headed the project's animation production team. A crucial component of the team's process was the development of special rigs. "We like to describe these rigs as everything in the kitchen sink and more," explains Rutland. "We tried to put everything we could into these rigs and then develop new technologies that we needed to guide the effects." The animation team manipulated previs models to build performances. Directing animator Allison Rutland built many custom rigs for the film.