Most of James Cameron’s space epic, Avatar, was shot on a performance-capture stage, known as the volume, in Playa Vista, Calif. The volume was rimmed by 120 stationary video cameras, which could record the movements of all actors at once in 3D, with submillimeter precision. Data from the cameras was streamed into Autodesk software, which translates actors’ movements into digital characters in real time within a low-resolution computer-generated environment. So riding a fake banshee mockup onstage instantly translated to CG footage. Multiple cameramen were used on set for reference video, but because the volume essentially captures performances from every angle at once, Cameron could digitally render whatever angles and shots he wanted after the performance, adjusting the camera movements while viewing playback.



Like many actors in Avatar, Zoë Saldana plays a fully computer-generated character, Na’vi princess Neytiri. To map her movements to her digital doppelgänger, Saldana wore a motion-capture bodysuit with reference markers and stripes. She also wore a head rig designed by Cameron that aimed a small video camera at her face. That camera tracked green ink dots, painted on Saldana’s face, throughout the scene, giving Cameron closeup-level detail of changes in expression to map to Neytiri’s CG face.



To shoot a scene within a totally CG world, Cameron had virtual production supervisor Glenn Derry rig up augmented-reality cameras. Cameron could watch from the sidelines as his actors’ performances were instantly mapped to their CG characters and displayed via an on-set screen. Or he could use a portable, motion-tracking virtual camera to walk through the volume and view the CG environment of the movie on its LCD screen.



To transition from the CG produced on set to the photorealistic world of the finished movie, Cameron sent his rough footage to Weta Digital in New Zealand. There, special-effects programmers used a facial solve program and facial action coding to translate the actors’ every minute muscle movement—blinks, twitches, frowns—to believable expressions on the faces of Pandora’s aliens.