kontext_depth_pose

v 1.1

added character reference, not sure if it works, :p

you might need to modify the prompt to discribe the reference character for it to work.

the success rate is low T-T

i ran 60 poses, 10 times for each pose, and get some results.

v 1.0

this wf used depth map to control output pose, success rate of complex pose is low.