Inference without --drv_pose

[siriuswi]

When I try to do video-driven inference without the --drv_pose parameter, the generated video is still not static, and it seems that the motion of the generated video is the same with the motion of data/raw/May_5s.mp4. Could you please tell me the reason? Is there something wrong in the code or my understanding about the word "static" is not accurate? Thank you very much！