KinecTris: Tetris + Kinect + Face Tracking + Gestures

Posted by Dennis on Dec 9, 2010 in 3D, Kinect9 comments

Here’s another Kinect experiment. This time a Kinect make-over of the best game ever, Tetris.

OpenCV is used for facial recognition. The depth value is then retrieved from the depth buffer by using the center point. The camera changes according to the face’s position. Simple gestures are recognised and used to control the position & rotation of the blocks. The face’s z-position is also used as a depth buffer threshold to filter out the face and everything behind it. Irrlicht is used for the 3D stuff.

This is not a fully functional Tetris game, just a demo. I might make a full game if there’s demand for it :-)



Tags: , , , , , ,


9 comments

» Comments RSS Feed
  1. [...] See Face Tracking Tetris with Kinect Leave a Comment LikeBe the first to like this post.Leave a Comment [...]

  2. Surely there’s a demand for this! I think loads of people are waiting to play with headtracking ever since they first saw johnny lee’s wiimotehacks.

  3. This is amazing! For me this kind of headtracking is the one of the best uses of Kinect. Even if you don’t want to make a full game, pleeease upload the demo so I can try it out! :)

    Well done!

  4. Good stuff Dennis! Really creative stuff. Linking the cool stuff together again. I am curious how the sensor deals with contrast in the web cam vision.
    But nice to see that you are still pushing the limits :-)

  5. Hi Dennis,

    Great hack, I would be really interested to see you code for face tracking with the Kinect. Are you just using the webcam on the Kinect to track?

    Cheers,
    Mark.

  6. Are you using OpenNI/NITE or OpenKinect?

  7. Hi Joan,
    I’ve used OpenKinect for this example.

  8. Hi Dennis,

    Users have asked me if it were possible to use Kinect for face-tracking in games (see http://facetracknoir.sourceforge.net/home/default.htm). I have not studied the Kinect interface, but obviously you have?!

    A few questions:
    - Is it right, that you must be 1.5 mtr away from the device?
    - You are using OpenCV. That suggests the Kinect has no internal functions to get the rotations/translations of your head? Are you not using the Kinect as an expensive web-cam that way? The CPU-load of the PC may suffer from that?

    Thanks in advance,

    Wim

  9. It would be great to see the source code – at least for the head tracking feature.

Leave a comment