problem with skeleton markers in pi-robot-ros-pkg with kinect

edit

Hi:

Has anyone used the above package with ROS? I ran roslaunch skeleton_markers markers_from_tf.launch which tracks using the openni_tracker and set my rviz config to markers_from_tf.vcg. If I use a fixed frame relative to one of the body points, say head or torso, I see the markers but not always all of them. It's like rviz shifted the image seen on the screen down or maybe its a zoom..

Anyway if I set fixed frame to the camera, ie. skeleton_openni_depth_frame, then I don't see the markers any-more. If I come real close to the camera I see the markers but they are all jumbled up and don't make any sense. Could there be a problem with the transform from the camera to the 2d representation?

Thanks in advance.