forums.naturalpoint.com

Hello!

I've been using the three-marker wand to calibrate my set of cameras everytime. But I've seen tutorial videos where it's used the single-marker wand. I wonder which is the difference between three-marker and single-marker wand. I know that the three-marker wand makes a more detailed calibration but, which kind of calculations makes the software with the aid of the wand?

I would like to learn more about how the calibration works.

Thank you!

The single marker wand was used with our older method of multi-mamera calibration which required more overlap of the cameras and was not as accurate. The new calibration allows greater freedom of camera placement and better use of capture volume space.

Both calibration methods calculate the intrinsic (camera and lens properties) and extrinsic (camera position and orientation).

http://en.wikipedia.org/wiki/Camera_res ... parameters

http://en.wikipedia.org/wiki/Camera_res ... parameters

Thank you for the information. I'm searching some documents where they explain the multicamera calibration. I've read that the fully self-calibration for virtual environments is made with a bright spot. Is this bright spot equivalent to the three or single-marker wand, in order to calibrate the set of cameras? In that document they say that you have to wave the bright spot through the capture volume, so I guess it's the same procedure.

Thank you!

Sure, It is a similar concept.

hint: In an ideal rectilinear imaging system (like a pinhole camera or a virtual camera) a straight line in 3d space makes a straight line in viewspace.

hint: In a non-ideal rectilinear imaging system (like a camera with a lense) there is often lense distortion, that results in straight lines in 3d space becoming bent in view space.

hint: When pushing two line segments of an unknown length through a projection matrix, their resulting lengths in viewspace are unknown if you don't know the field of view. BUT, you can be fairly certain the longer one will still be longer than the shorter one within reason. Technically, a really bad lense distortion, or optical center shift, or a very oblique angle could ruin this. But statistically, you can eliminate these bad samples.

Thank you for the information again. Brad, about lens distorsion, would that be the reason why the distance between markers is different in the three-marker wand?

That would be a reasonable assumption... but me thinks I've said too much already...