Issue link: https://digital.copcomm.com/i/691968
m ay . j u n e 2 0 1 6 c g w 1 5 created the rig that animators used to pose Hank and his seven tentacles. One of the first reference videos Talbot looked at to determine how an octopus would move was of a mutable creature sliding into a bottle. "It terrified me," he says. "Not the look, but how it moved over a rough surface; the way the flesh rolled." First, the team created a con- trol that animators could use to set the depth of the webbing between the tentacles, and de- veloped a new curve function. "Aer a lot of work, we had something for animation to play with," Talbot says. "But, we didn't have the sliding motion we needed." For that, the rigging crew turned to the simulation department for help. There are 350 suckers on Hank's tenta- cles, and each sucker needed to squish, stick, and pull away in a believable manner. Solid mechanics came to the rescue. "There's a whole branch of engineering that, given a force applied to an object, asks how are all the points in that object moved, displaced, and stressed," says John Halsted, supervising technical director. "We borrowed approaches from that field of science to figure out how Hank moves." Thus, each of Hank's suckers has a little tetrahe- dral mesh that provides input into a customized simulator. The simulator produces the squashing and sticking; it de- forms the little sucker based on applied forces. "The simulator is a heav- ily customized version of Physbam," Halsted says. "It uses finite-element analysis (FEA) to have a solid, elastic material behave like muscles or flesh. We use the tetrahedral mesh to do the volume-based jiggle and solve it with FEA. To get the skin sliding, we used a triangle mesh over his body." Once the team managed to create one sucker that would squish, stick, and pull away, they could propagate the method to all the suckers. Then, they needed to make the tentacles look so and fleshy as it squished along. "A separate skin simulation allowed the skin to move along the same plane as the outer sur- face of the body," Halsted says. "It doesn't break the silhouette. It just slides along." The techniques used to create Hank's simulation are likely to result in a SIGGRAPH talk this year. W A T E R The second challenge for the technical crew centered on wa- ter. When Pixar created oceans of water for Finding Nemo 13 years ago, 3D water simulation was in its infancy. For this show, the team concentrated its ef- forts on water compositing. "The way we approach effects is that if the characters are in a large body of water, we do a smaller domain where they are playing," Halsted says, "a play area where they can jump around and break the water apart. So we worked on tech- niques for this film in which we take a large body of water usually shaped with procedural systems, and composite in the 3D simula- tion from the play area such that the boundaries are seamless." For fluid simulations, the team used Side Effects So- ware's Houdini; for lighting, The Foundry's Katana; for rendering, RenderMan RIS. "We came up with a new system for doing the water composite that makes heavy use of an implicit field API to generate the surfaces," Halsted explains. "We also worked on water shading to make sure the water looked as good as it could. And we made all this work within the new renderer." The fish in Finding Nemo spent most of the film in the ocean. Finding Dory's setting in the Marine Life Institute meant that oen the fish in this film were in the aquarium's glass tanks. The lighting crew, led by Ian Megibben, put the new renderer through its paces for test shots with Dory in a fish tank and Hank staring in at her from outside. "There's an inherent quality to the way water bends, re- fracts, magnifies, and scatters light," Megibben says. "Water inside a glass tank is like a fun house. There's magnification, distortion, and reflections. When the camera is from Do- ry's point of view, we see reflec- tions that create a boundary between her and Hank." Because Finding Dory is the first film for which Pixar used the RIS architecture, they employed Katana to author new shaders. "It was a huge change," says CTO Steve May. "It affected all the back-end departments – lighting, rendering, shading. On Dino, we were still doing spherical harmonics and special shadow maps. Now, we're going to raytracing. "We felt like we were a little behind the curve," May contin- ues. "But the good thing is that DESTINY, A NEARSIGHTED WHALE SHARK, MEETS LITTLE DORY.