forums.naturalpoint.com

Hi,

I am new to using OptiTrack. I am interested in capturing the motion of shafts with a very high accuracy (1-3mm) within a relatively small volume (2'x2'x2'). I currently have three V100 cameras and the Rigid Body toolkit. I would be interested to hear any suggestions for improving accuracy for this situation. Specifically, I am interested in:

1) How far should the cameras be from the capture volume and how far from one another should they be?
2) Are there differences in accuracy between the 7/16" reflective markers and the 3/16" facial makers? Are spherical markers preferred or would a strip of reflective tape wrapped around the shaft provide similar accuracy?
3) When calibrating the cameras, what are your recommendations for threshold, exposure, and illumination for optimizing accuracy?
4) When waving the calibration wand in the capture volume, is there a preferred pattern? I have been trying a helical motion as suggested on these forums for the ARENA setup.
5) Following the calibration calculation, if standard or mean error is high, is there something specific that can be done to improve each of these?

Any other tips you might have for improving accuracy would be extremely helpful. Thanks.

1) Try to have the cameras at least a couple feet apart. Three to four feet apart may be optimal for your volume with the cameras about the same distance back from the perimeter of the capture volume. For best calibration and tracking results, avoid placing the cameras all in the same plane. Instead position the cameras at different angles, for example your three cameras could be placed in an arc all facing the center of the tracking volume.

Also try to make sure all the cameras are pointed toward the same region in space. Placing a reference marker in the center of your capture volume to assist in aiming the cameras is an easy way to do this during setup. Try to center each camera's FOV on the marker.

Since every setup is going to have different needs and restrictions, we really recommend experimenting to see what works best for you.

2) Larger markers will typically show up to the camera with more pixels than smaller ones, this allows the marker's center of mass to be calculated with more resolution.

A strip of reflective material wrapped around the shaft would be less ideal. Also keep in mind that rigid body tracking works *much* better if the 3+ markers are not co-linear. Forming a triangle instead of a line will allow the 6DOF solver to work properly.

3) For V100 cameras illumination should be left on 15 (strobe), threshold should be left around the default range of ~140. Since your cameras will be close to the objects they are tracking you may want to reduce the exposure to between 5-20, this will lower the illumination (to avoid reflections and blooming).

4 & 5) When wanding, try to evenly cover the entire 3D region of the capture volume, also get as much coverage for each camera as possible. If you don't get good calibration results try a different wanding technique, make sure the sync cables are connected properly and check the camera aiming as mentioned in #1. Also make sure masking has been properly performed to avoid picking up unwanted objects during the wanding process.