All of the creatures for the story — including the Troll, Balrog, mumakils and oliphaunts, and the charmingly ancient Treebeard — began as design maquettes, before being rendered as digital models. The models were then fitted with a skeleton, a muscle rig was built around the skeleton to drive the flesh, and then the characters were given texturing and lighting for the final scene.

Eric Saindon, CG creature supervisor for The Lord of the Rings trilogy.

Eric Saindon served as a creature supervisor for The Two Towers, mostly working to set up the muscle, skinning and hair on Gollum. Saindon comments, “Our muscle and skinning system [at Weta], the way the muscles flex and keep their volume, the way the skin actually moves on top of the surface, is a little unique compared to most of the other systems I have seen.” The muscle code for the creatures was written entirely in-house as a plug-in for Maya.

Saindon recalls how Gollum, who appeared in The Fellowship of the Ring as little more than skin on bone, had to be redesigned for The Two Towers: “It was a bit of a daunting task to make Gollum look real. In the first film, we faked it quite a bit to get through the two or three shots we had. You hardly had to see him. But we really couldn’t do that in the second film.”

Originally from Maine, USA, Saindon studied architecture in Washington state, before later moving down to California to work on environments and spaceships for the feature film Star Trek: Insurrection. Moving from his role as technical director into animation set-up, he eventually found himself working on the fantastical creatures of The Lord of the Rings.

Among his other duties, Saindon also helped to establish the pipeline for bringing motion-capture data from the actors’ movements onto the digital puppets. Nuance, created by Giant Studios, is the modification software that Weta used for The Two Towers in editing their motion-capture data. With this foundation, animators would then change, edit or in some cases completely redo a character’s performance.

A concept painting for Helm's Deep (top), and the composited scene (bottom) using Nothing Real's Shake and Weta's proprietary Massive software.

“We didn’t use mocap data on all the shots,” Saindon explains, “and for those that we did, typically, we didn’t use straight mocap data.” Prior to skinning and texturing, he says, “Our motion edit group took the data, cleaned it up, got rid of the pops and slips and other artifacts of the process… That information was put on an IK puppet, and then the animators changed the performance to work better in the environment of the scene…or sometimes they would delete entire chunks of the data and re-animate it from scratch.”