Duration: 0:41

283 views

14

This is a short demonstration of an AI I created. It was trained on ~40 minutes of footage of the VTuber Korone translated into black and white to pick out the details of her poses and automatically create sliders to ajust them.

The approach was principle component analysis on about 1600 frames of one of Korone's videos, which were used to train a neural network AI via backpropagation. For more information, I recommend codeparade's video on generating faces.

This video is made for the purposes of education as it demonstrates what can be accomplished with artificial intelligence, and is thus subject to fair use in the USA.