post processing a ROS bag - how to parse the point clouds after tf?

edit

Hi all, I have a robotic system based on ROS, in which I record the data (PointCloud2 messages) from several different sensors, all calibrated using static tf. While in rviz the data seems calibrated, when I process each topic (of the different sensors) from an existing rosbag using python, I believe the point clouds in the different messages are not calibrated.

Is this true, and if so how to what to do so the output will be calibrated as shown in rviz?

Thanks