libgazebo_ros_openni_kinect.so depth pointcloud wrong tf

Hello, I am new to ROS. I am trying to simulate a robot in gazebo witch works fine for now, but I have a problem with simulating a kinect camera. The simulated data seems to be right, but when I try to display the results in rviz the transform is not correct. Can anyone give me a hint where my coding went wrong? Gazebo: 6.1.0 Ros: Indigo Rviz: 1.11.8  model.sdf

0.29 0.0 1.068 0.05 0.535 0.130.10.1 0.1 0.10 0 0 1.57 0 1.57model://b45/meshes/asus.dae1true1.3962634640480R8G8B80.1100true30.0cameraLeft/CameraLeft_LinkCameraLeft/rgb/image_rawCameraLeft/depth/image_rawCameraLeft/depth/pointsCameraLeft/rgb/camera_infoCameraLeft/depth/camera_info0.83.5

robot.urdf