Inaccurate Head and Neck movement

Post Reply
benjcooley
Posts: 1
Joined: Tue Mar 21, 2017 10:13 am

Inaccurate Head and Neck movement

Post by benjcooley » Tue Mar 21, 2017 10:21 am

We've been using Motive for a while, and we've noticed repeatedly that the skeletal rotations for Head and Neck seem to be inaccurately inferred from the marker data.

Specifically, when the head rotates left and right, an incorrect orthagonal rotation on the neck joint causes the entire neck to seem to "swing" left or right. We are constantly having to go into the mocap data and correct this by removing the lateral rotation on the neck so that the normal and expected rotation looks correct when applied to the character.

Likewise with head rotations. It seems as though "look up" and "look down" rotations are greatly exagerated, and are causing additional unwanted rotations in the neck joint which - when directly mapped to physically realistic rigs, makes the head appear to nod downwards in physically unrealistic ways.

My question: is this a well known problem in Motive in inferring accurate motion on head and neck joints? Have we miscalibrated in some way (we've spent time with the actors and software to try to figure out what is wrong, but have not been able to identify it).

Having to correct this issue, or to build custom rigs which can correctly constrain unwanted and unrealstic head and neck mostion is becoming a problem for our workflow.

Any information that anyone might have on this issue would be greatly appreciated.

steven.andrews
NaturalPoint Employee
NaturalPoint Employee
Posts: 500
Joined: Mon Jan 19, 2015 11:52 am

Re: Inaccurate Head and Neck movement

Post by steven.andrews » Fri Mar 24, 2017 2:27 pm

Hello benjcooley,

Thank you for reaching out to the community regarding your results.

To begin, we should just note that the skeleton representations in Motive are not necessarily an accurate representation of the actual skeleton, and there may be slight deviations from actual joint positions and rotations.

For purposes such as animation, the skeleton data is usually close enough to reality to represent a performance. I have not seen a great deal of error on the head in normal setups, but it is certainly possible your head may be rotating off the wrong pivot point. This could be made worse if marker placement is off.

It would be best if you could reach out to us directly, and share a Take (.tak) exhibiting the head movement you are observing. This would allow us to see if your results are far off from what is expected. If you happen to have any reference photos of the performer and the marker placement on the performer, this would also be very useful.
You can reach out to us at help.naturalpoint.com for further troubleshooting.

Best,
Steven
--
Steven Andrews
OptiTrack | Senior Customer Support Engineer

Post Reply