Dance is the hidden language of the soul and body that only a few can speak through their feet. And if you are one of those who has two left feet, then the new AI technology can solve your problem.
Now, you can show your dance moves like Beyonce and Michael Jackson with the new AI tool designed by researchers at UC Berkeley; it is being known as deepfake for dancing.
Deepfake is an AI-based human image blending technique. It is utilized in merging and superimposing existing images and videos on the original photos and videos. Moreover, deepfake face swap was banned on many sites such as Reddit Pornhub and Twitter in 2017.
UC Berkeley researchers published this new found-out on arXiV and titled it as “Everybody Dance Now”. The research paper proposes a measure to transfer motion between human subjects in diverse videos. There could be two videos, target and source to make the subject dance like a ballerina or perform martial art kicks.
To be able to provide motion in the video through frame-by-frame, one will have to understand the mapping of images between two people. Thus, the aim is to explore an image-to-image transfer between the source and target videos sets.
The study states, “To accomplish this task, we divide our pipeline into three stages – pose detection, global pose normalization, and mapping from normalized pose stick figures to the target subject.”
The creation of the video requires some clever work as there can be some limitations in producing the target video. Subroutines are put to practice to smooth out the dance moves. Also, the dancer is supposed to wear body-fighting clothes as the network won’t be able to model loose ones correctly.
As described by the researchers, amidst the challenges faced while creating AI dance video, their method will be able to produce compelling videos given a variety of inputs.