Kyle_butler wrote:
Do you mind using dropbox?
Absolutely not. Just mentioned another way of sharing FYI.
Kyle_butler wrote:
I've uploaded both of the files to it already.
OK, got them.
Another bad thing about lights off is that other cameras are not visible, so it is not easy to figure aout how they are located. But through the motions of light I've managed to get that it's somewhat like a full circle in clockwise direction.
I've processed the video with "Auto detect initial camera positions" off and "Full circle clockwise" as initial cameras placement. Got quite good results as I can see that red leds visible in some cameras matched with positions of the opposite cameras. The results are very close to what you've got in your project after I have slightly adjusted the ground points to be the lowest marker positions in each of approaches to the floor, and set the scale in both projects to have Camera 1 height equal to 1.800m.
I've attached both project files in ZIP archive. Yours
2013-04-11_10-26-38.iPiMocap with corrected ground points, and the one I've processed myself
2013-04-11_10-26-38.new.iPiMocap.
I can say that the quality of video itself is sufficient for performing the calibration. So, what you consider wrong with your calibration results? Camera heights differ considerably from the measured values after scaling? Is it fixed by the more accurate selection of ground points?