Hello there my friends, now the idea is based on a problem ofcourse. anyway, same as you set 3-5 points as ground, why not set 3-5 t-poses as a colour guide for each region of the work space?

While i have been tracking dozens of videos, colour was the major problem in the rig dislocating from me (actor), i mean if my rig dislocate, all i have to do is detect actor colours and re-track.

The reason is; as much as i tried to keep the colour uniform in all parts of the workspace, i was able to get 75% uniform and 25% still darker than other parts. i used a light source behind each camera (3), and another light source for support (each source have like 4-8 neon bulbs).

Trust me, i tried so many combinations of light sources to get maximum uniformity. So, it's like setting 3-5 points of colour and let an algorithm create the gradient, but which colour to use from the gradient space is based on the position of the hip or pivot of a bone.

This should cover any combination of light sources even low light, should even give much better tracking performance and higher resolution and less jitters!

Just an idea to share, hope it works out!