Video encoding for OpenNI topics to record them to a rosbag and replay through reconstruction of point clouds.
Includes libav_image_transport by Dominique Hunziker: https://github.com/rapyuta/libav_image_transport.
Before doing catkin_make on the containing workspace, you need to fetch the submodules one time
git submodule initgit submodule update
within the git repository.
To put the compression gains of this library in context, here is a comparison to the default ROS image compression and the uncompressed images.
Both the ffv1 codec and the theora codec takes up quite a lot of CPU when compressing the OpenNI streams at 30 fps. The following CPU usages were measured on my Core i7 laptop.
