4th axis (Z axis) stabilizer

I recently purchased a z axis stabilizer to try to reduce the amount of vertical oscillation when walking but I got mixed result. As it was hard to compare before and after as I don't have 2 gimbal to lock on the same tripod I decided to plot the input of the stabilizer (the stabilizer handle) and the output of the stabilizer (The gimbal handle).

So I tried to reproduce a motion similar to walking but on place and using after affect I tracked both handles position in space.

I've exported the data into excel. Here I looked only at the data in Y (vertical) but there is also an effect on the X (forward axis). But the impact in the X axis doesn't impact the video stabilization as much so I focused on the Y axis only.

From there it was pretty hard to evaluate the benefit. The output seem to have a little bit less motion in it. Note that Y out is the gimbal handle position and Y in is the stabilizer handle position.

So let's do a frequency analysis...

The low frequencies (left of the graph) are less important as they results into smooth motion. Starting from 0.5Hz (One cycle per 2 sec) it starts to get noticeable on the video. We can see that the stabilizer does a good job above 1.5Hz, but it has a natural frequency at 0.9Hz which get amplified when the motion in continuous. (Which is frequently the case when walking at constant speed)

Note that the natural frequency depends on the weight and spring adjustment, so with a phone installed it will vary slightly.

Now let's see the difference between the two.. (Lower values are better with the stabilizer)

That is not a clear winner indeed. But once again it is pretty hard to evaluate to I computed the global attenuation for all frequency above 0.5Hz. I got an overall attenuation of the Y frequencies of -2.14dB. In other words, the output Y axis motion is 78.2% of the input Y axis motion or a reduction of 21.8%. In term of video motion this is not a lot but it helps a bit.