NVidia shows off their latest tech in video synthesis, allowing a single image of a person to turn into a dancing star!

The paper recently released by the researchers really demonstrates some astonishing improvements in deep learning and artificial intelligence. The video shows two different achievements.

The first takes a still image of a person, or even sculpture or picture, and animates it according to another video animation sequence. Some examples shown are a single person image dancing according to a given dance sequence, a sculpture image dancing according to a given dancing person video, and a single person portrait image (or painting), talking according to another talking person example video.

The other example demonstrates scene transfer, where a single image of a scene describes the feature or style, and that is that transferred onto another video animation, trying to imitate the initial scene.

Honestly, the video speaks for itself, enjoy.