Quote Originally Posted by kirtley24 View Post
Not quite true - should be 6 x natural frequency. So gait at 120 steps/min would be 6Hz, but 60 steps/min could be filtered at 3Hz and 180 steps/min would need to be 9Hz.
There are two factors that need consideration - I agree with Julius that it depends on what do you want to achieve, however, when sampling the motion I think that it's important to sample much faster so that you can review the raw data for any issues prior to filtering it. An oscillating marker or sensor can generate signals that distort the data but are completely lost if the sample rate is low. So filtering at 6Hz is fine, but sample much faster to guarantee the data quality.

Many years ago was once asked to diagnose problems in a gait lab that was generating odd joint moments and powers in the clinical reports, with the customer unable to figure out why. It turned out that the force plates were badly installed and were ringing when the subject stepped on or even near to a force plate, but since the data was being sampled at 15Hz and filtered at 6Hz, nobody could see this.