The developers from UC Berkeley in the US have created a network for those who can’t dance, but wants to impress friends videos of the dances.
One algorithm creates a virtual skeleton with a map of poses, and the other two to create the full picture, transferring them to the right person, with realistic facial expression of the object as the movement of the virtual body.
Technology is still far from fully cooked. The video can be seen trembling, blurry areas, and even lost body parts. However, the result is already impressive.
The scope of technology is understandable, although it leaves some ethical questions open. For example, you can “make” unnecessarily sandwiched fellow workers to dance the shuffle under elektrokhaus.