How to ensure odometry messages are published at same rate as input point cloud?

[ahmedadamji]

My input pointcloud in a bag file has 1104 messages, however I am only able to yield around 900 odometry messages, which are for the whole trajectory but not for each instance of the input point cloud, which in my case is critical for evaluation. Could you please guide me why this may be occuring and how I can solve this?
I tried to change the queue size in the hdl_localization_nodelet, changed the ndt_neighbor_search_method to DIRECT1, yet, which has only helped marginally.
Thank you

[wienans]

Did you try to run the bag file with a slower rate? In this case you would ensure that it is not a problem with Computational time