Hello iPi.

During our motion capture test with PS-Eyes and Dual kinect 2 we started wondering what the difference is between the board calibration and the light calibration. As we have tried both of them multiple times now our conclusion is that the board calibration is very easy to perform aswell getting a perfect result (~ 2 cm error). Getting perfect calibration with the light method (using a maglite) seems not that easy. When we finally got the calibration to be perfect, the margin of error was even better than with the board calibration.

We have been reading on the internet and this forum, and it seems like the light calibration should be more accurate - is this true?
Also, we have noticed on the "new" iPi user guide webpage (http://docs.ipisoft.com/Main_Page) that you do not refer to any kind of board calibration anymore, even tho it is still an option in the iPi Mocap 3.3.9.205 version.

1. Are you able to explain the difference between the light calibration and the board calibration in terms of accuracy?

Greetings

Rephill

Hi
Actually, new docs contains a small footnote reference to the plane calibration method. However, we do recommend the light method because from our experience it's easier for recording and gives more stable results. Also, this method is suitable for any number of depth sensors while the plane method only fits dual config.
We do not state that the light method is much more accurate, but it is a bit. Simply because computations involve less variables. In the light method, we only need to find a center of light spot on color picture and determine its position from corresponding point on the depth map. While in the plane method a plane is averaged from a large selection of depth map points and a board's corner points are found. This makes the latter method more sensitive to the noise in depth data.

As in your experience, I find the board method easier too. However, I do think the light method is more accurate and, as noted above, it's necessary if you want to record with three or more Kinect sensors.

The tricky bit with the light method is that Mocap Studio must be able to see it at all times, and if you have large reflective surfaces in the room or bright light coming in from a window, that light can become harder to track. Basically, you need to make sure the light is clearly visible and distinguishable from other lights or reflections that may be in the room. Check the RGB video for anything that might cause the tracker to get confused. For example, lots of glass on the walls, a large mirror, or a very shiny floor might potentially cause a problem.

_________________
Greenlaw
Artist/Partner - Little Green Dog | My Demo Reels (2013,) (2015,) (2017,) and (2019)

Watch a one minute excerpt on Vimeo now!

Thank you for the replies.
We will perform some tests to see if there is a difference between using a scene from a board calibration compared to the light calibration. Both of course needs to be perfect calibration in order to do a proper comparison.

Rephill