Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!usc!zaphod.mps.ohio-state.edu!uakari.primate.wisc.edu!dali.cs.montana.edu!milton!hughes@volcano.Berkeley.EDU From: hughes@volcano.Berkeley.EDU (Eric Hughes) Newsgroups: sci.virtual-worlds Subject: Re: Using bare hands Message-ID: Date: 8 Nov 90 02:49:38 GMT References: <9319@milton.u.washington.edu> Sender: hlab@milton.u.washington.edu Organization: ucb Lines: 36 Approved: hitl@hardy.u.washington.edu In article <9319@milton.u.washington.edu> gourdol@imag.imag.fr (Gourdol Arnaud) asks what camera sensing accomplishes better than datagloves. The dataglove I have used (one of VPL's) cannot detect the following gestures, all of which cameras could. 1) fingertip touches, e.g. make an OK sign. 2) more generally, contact of any sort. 3) the articulation of the bones in the palm (could somebody supply the name?) in their motion from the wrist. As a special case, there is the position of the base thumb knuckle relative to the back of the hand. 4) quick hand motions. I would love to be able to draw letters in midair and have them appear in the space; generalize to Chinese characters. Another gesture I would like to have is "wiggle your pinky." The sampling limitation of the Polhemus is part of the problem; a quick 90 degree turn and back may not take. The rest may or may not be "a simple matter of software." I don't know. 5) the other angular degree of freedom in the articulation of the finger from the palm. Currently only one angle is sensed. An example of a gesture is the "whoop-de-do" motion of a straight forefinger tracing out a circle with the tip. Here is a litmus test of the power of any hand sensor: can it reliably detect the sign language alphabet at the rate of four characters per second? Eric Hughes hughes@ocf.berkeley.edu