accuracy question

[inzane]

Hi,

I've read some papers where people have used OptiTrack for tracking of an object in 3-space. I was wondering what the actual accuracy resolution/accuracy specs are for this system. So for instance, can you track a point to .001 inches, or is it more loose than that?

Thanks in advance!

[VincentG]

The 3D location of markers can be resolved with millimeter accuracy and resolution depending on capture volume size and camera configuration. Increasing the number of cameras can help improve the tracking performance if needed.

[Jim]

Just to add to what Vince says; the marker resolution can be sub millimeter for sure, down to .01mm, but accuracy is a different matter, and greatly depends on the number of cameras seeing a marker, the distance to the marker, and the size of the marker, not to mention needing a good calibration result.

We have seen accuracy around 1mm absolute, if not better for most capture volumes. If you post your ideal setup, we can be more definitive.

[inzane]

Fair enough. What we're doing is trying to record the position and orientation of a small rod (say 1 inch long) in a bounded space of 18 X 18 X 18 inches. The cameras wouldn't be more than 5 feet away from the subject. Basically we need to verify the readings of a magnetic POM with 6DOF. I was thinking 3 or 4 markers, but I'm not sure of how many cameras would be required.

[Jim]

This sounds like a setup where we could do better than 1mm. You can see the noise in the data as it is streamed out, so you can see the certainty it is reporting with.

I would suggest 4 cameras with 3 markers. We tend to work better with more cameras, but 3 markers / rigid body.

I would try to spread the 3 markers out more than the 1" of course, to make the triangle more accurate and so the markers aren't occluding eachother. The bigger you make the triangle, the more accurate it will be, 6" could be good, if I am envisioning the volume correctly.

[inzane]

That sounds about right. Thanks for your help guys.