Hmmm (Score 1)

Consider this.... The software must analyze the video from the webcam with probably neural net technology or something akin... While this is going on the system must render the 3d world on the users screen... Then bear in mind that webcams are low res... The bottom line : this is not feasible. It would be easier, less computationaly intensive and cheap to just use ultrasound with sensors attached to hands elbows and whatnot...