Transformers 2 Nearly Broke ILM

The rendering demands for Transformers: Revenge of the Fallen were such that some ILM computers didn't not survive according to a new article from SciFi Wire. Sections below, full article here.

Visual-effects supervisor Scott Farrar was in charge of turning Industrial Light & Magic's computers up to 11 to create the new characters and told reporters that the sequel features 40 new characters. That and the increased resolution of the characters for new IMAX footage nearly exhausted ILM's render farms: After one hard night of rendering computer-generated footage, some of the hardware actually exploded. "We did, we lost some machinery that night," Farrar said in a press conference on last week in Beverly Hills, Calif. "Little puffs of smoke, just like in the movie."

The largest sequence in Revenge of the Fallen was also the biggest in ILM history: the climax in which Devastator tears apart one of the great pyramids in Egypt. "We're trying to hit new levels of realism in every single thing we do, whether it's the render of the robot or the physical environment that they're reacting with," Farrar said. "It's just like upping the game on every level, so it was a pretty complicated show."

To give you some sense of just how big Devastator is, Farrar said that Optimus Prime has 10,000 moving parts. The computer algorithms actually manipulate each part to go from truck to standing robot. Well, Devastator is made up of upwards of 80,000 parts. The only thing that saves time is the camera position. The animators only have to transform the parts that are visible on screen.

"Every shot is dressed to camera," Farrar said. "We have a lot of moving parts and a lot of pieces that are all finished up, but every single time that we set up a new camera position, the cameras swirls around to the back, and doggone it, there are some pieces that are unfinished. We have to repaint them and get them so they can be animated. They only move if we need them to move, so it's a logarithmic jump to try and get all those pieces to move, and it's all up to the animators, frankly, to lay down the movement first. We tried to free it up to be creative."

"Everything that we do in our world is all about the light," Farrar said. "It's not just building the robot, but it's about how it commingles with all the light sources. It might be ambient. If they're really deep in the water, how much light do we give them? How much internal lighting should they have? We had a lot of deep-sea underwater research photos that we looked at, and we sort of gleaned from that. How clear do we want to be? How much plankton and spinaci, we called it, floating in front of the camera do we want? All these little tricks to try and make you believe you're really underwater, we have to employ."