InfoCentral wrote:
I think the new features just released in the
Kinect SDK 1.5 maybe a source of delay. This should greatly speed up the head and hand tracking implementation. The facial mocap and other features would probably take more time.
Currently we are not using Kinect SDK skeletal tracking. We only use Microsoft drivers to get depth data from Kinect. Kinect SDK skeletal tracking is very limited compared to our software. There are simple poses (like sitting or bending a knee) that cannot be tracked with Kinect SDK. So there is little sense in using it for animation. Of course, Kinect SDK tracking is very fast (realtime), but in animation accuracy is more important.
Hand tracking and head tracking in Kinect SDK is rudimentary and is hardly useful for full-body animation. Of course, Microsoft has pretty good head tracking at close range (1.0-1.5 meters), but this is not what we need for full-body mocap. At 3.0-3.5 meters Kinect SDK's head tracking works very unreliably.
InfoCentral wrote:
The facial mocap and other features would probably take more time.
We are not working on facial mocap as of yet. Part of the reason for this is the fact that other companies like Image Metrics (and now their spin-off Faceware) are doing a very good job with markerless facial mocap. We feel we should concentrate on full-body markerless mocap for now, because that is where the challenge is.