Love this work, and easy to follow! But one question...

[malcolmgarner-movement]

Hi @ChrisWu1997 ,

As a complete neophyte at Ai and machine learning models, having this to play around with has been very enlightening, and I dare say, enjoyable! Watching skeletons do what we want is immensely pleasing!

This is more of a question rather than an issue however: what is the process called when you map the actions of the human in the driving video, onto the human in the reference video with the bodies and the background?

Is it just known as Video performance cloning? Or is there a more technically specific term for it?

I'd love to get my teeth into that topic myself.

Much obliged!

[ChrisWu1997]

Hi, thanks for taking interest in our work!

We call it performance cloning mainly because we used the method in this work Deep Video-Based Performance Cloning. I'm not sure is there a uniformly agreed technical term for it since I'm no longer working on this problem, but I can provide some other related papers that I know:

• Synthesizing Images of Humans in Unseen Poses
• Everybody Dance Now
• Video-to-Video Synthesis
• Neural Rendering and Reenactment of Human Actor Videos

You may start by reading these papers and following works if you want to get into this topic:)