A new artificial intelligence system that can generate a fake video clip from a single photo has been presented. Deepfakes or fake clips that make the subjects say and do things they haven’t usually require huge datasets. This has been one of the factors keeping the technology away from bad actors. The research has resulted in a system that first performs ‘ lengthy meta-learning on a large dataset of videos, and after that is able to frame few- and one-shot learning of neural talking head models of previously unseen people as adversarial training problems with high capacity generators and discriminators.
Crucially, the system is able to initialize the parameters of both the generator and the discriminator in a person-specific way, so that training can be based on just a few images and done quickly, despite the need to tune tens of millions of parameters. We show that such an approach is able to learn highly realistic and personalized talking head models of new people and even portrait paintings. The software can have fun applications like being portraits to life. But what many fear is this software will end up in the wrong hands and be used as a tool for political and social manipulation.