'Faceshift' Delivers Markerless Motion Capture Via Kinect

So it might not deliver the kind of sophistication in motion-capture that was utilised by Team Bondi for its ground-breaking crime caper L.A. Noire – that, after all, relied on a studio set-up with 32 high-def cameras tracking a single actor’s face – but new facial animation software ‘Faceshift’ does more than an adequate job in replicating such techniques, and it relies solely on Microsoft’s Kinect.

Though we’ve at times been frustrated with many of Kinect’s shortcomings when it comes to simple motion-tracking in-game, the camera continues to impress when the Windows version of the peripheral is put in front of developers. Faceshift, developed by a team from EPFL (École Polytechnique Fédérale de Lausanne, School of Computer and Communication Sciences Computer Graphics and Geometry Laboratory), is the very latest to catch our eye.

Using Kinect to capture the 3D data it needs to construct an accurate computerised model of the users’ face, the software does a terrific job in further mirroring every expression, facial twitch and movement, shown off with barely any noticeable delay on an on-screen avatar. Developers can improve the accuracy of the facial recognition even further in a post-processing stage.

According to the team behind it, the software tracks even the smallest emotions thanks by tracking 48 different 'blendshape' parameters, and can even track gaze and sense blinking. Currently being licensed to developers for $1500 a year, it's a snip for those studios looking to get decent-quality motion capture into its games, all without the hassle of a hi-tech, lab-like studio set-up.

Richard Birkett

Jason England

I am the freelance tech/gaming journalist, lover of dogs and pizza enthusiast. You can follow me on Twitter @MrJasonEngland.

http://stuff.tv/team/jason-england
Previous
Previous

Scientists Devise Armageddon-Like Nuclear Bomb Asteroid Plan

Next
Next

Researchers Create Walking Biological Robots Using Heart Cells