JARVIS-MoCap / JARVIS-AcquisitionTool

AcquisitionTool to record multi-camera recordings for the JARVIS 3D Markerless Pose Estimation Toolbox

Home Page:https://jarvis-mocap.github.io/jarvis-docs/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Strategy for calibration?

olinesn opened this issue · comments

Hello,

I'm having difficulty running calibration with my camera setup. I have 7 cameras, and two of them are pointed inward in such a way that they rarely / never see the charuco board at the same time. It seems like this is causing problems, but maybe it's related to how I'm setting up the pairs?

In my diagram, the two red circled cameras are the ones highlighted in the error. I used the blue circled camera as the primary, and so every pair is the blue circled cam and one of the other cameras.

image

Should I use a different strategy when setting up my inputs? I see that in the guide, one of the pairs is actually a set of three, but it's not clear to me which cameras should be selected for that, or if it's even necessary.

Thanks for your help!
Stefan

image (25)
image (27)

I think in this case using the 3-camera pair option is the best strategy. The idea behind it is that instead of directly calculating the position of the secondary relative to the primary camera, it first calculates both the position of a intermediate camera relative to the primary and the position of the secondary relative to the intermediate camera. For your specific setup if you want to calibrate camera e3v82aa with e3v833e as the primary, a good 3-camera pair would be e3v833e->e3v8394.

To set up a 3-camera pair you have to double click a camera pair in the "Camera Pairs" list. You will then get a pop-up window that lets you modify this pair.

I hope this helps, but feel free to ask if you need further clarification.

Hello,

I want to thank you for your constructive and specific comments. I've been able to succeed with calibration following both your instructions and actually making one more change: using a larger charuco board (11" wide instead of 7" wide). Not only is calibration succeeding more frequently, but it actually runs much much faster than before.

Thanks!