MotionScan Gives the Lie to the Uncanny Valley

This making-of video describes how Australian company Depth Analysis takes motion capture to a new level in MotionScan, a system they developed for Rockstar’s forthcoming detective game L.A. Noire. 32 scanning cameras are placed in pairs�presumably to take advantage of the parallax effect of stereoscopic vision uses�above and below the seated actor. There’s a lot of high-speed computing involved in knitting all of those scans together into a full-featured facial performance and flattening the files not only for cut scenes, but for in-game animation as well; many technical details are found in Gizmodo’s lavish report on the studio and its approach to the motion-capture problem. But this sticks out for me: the team’s question was, can we capture the look on a person’s face when he’s lying? That’s the limit for the uncanny valley effect, isn’t it? Detecting a lie, after all, is something our brains do effortlessly; but previous motion-capture for gaming has collected too little data to render the twitches, the shifts, the blankness in the eyes that comes with prevarication. Bringing lying to video games doesn’t only enrich cut scenes�it opens up a whole new range of challenges in the gameplay itself.