NVIDIA researchers published a new paper (PDF) that describes an alternative generator architecture for generative adversarial networks. This new architecture enables a neural network to create fake images of human faces that look extremely realistic to the untrained eye.
The neural network uses several input photos to create a completely new face, and it's tweakable thanks to the use of styles.
What’s particularly interesting is that these fake faces can also be easily customized. Nvidia’s engineers incorporated a method known as style transfer into their work, in which the characteristics of one image are blended with another. You might recognize the term from various image filters that are popular on apps like Prisma and Facebook in recent years, which can make your selfies look like an impressionist painting or a cubist work of art.