akanazawa/human_dynamics

about output parameters

Closed this issue · 2 comments

XihuaQiao commented

is there any chance I could get the rotation (global orientation) vectors and translation vectors of the model for each frame? And the rotation and translation variable of cameras?

XihuaQiao commented

Does the root orientation which refers to the mean of 'L Hip' and 'R Hip' joints represent global orientation?

jasonyzhang commented

Global rotation is the first rotation matrix in the "poses" key of the output dictionary.
We use a weak-perspective camera model (scale + translation), so you can approximate the global translation following #57, #59.