StyleGAN

machine learning to generate and customize realistic images

Screenshots

Hunter's comment

Generative Adversarial Networks, where the high-resolution human face images can be created by combining two sources or with other attributes without any human supervision.
There is no need to sit on graphic designing software for hours to create characters or images. Machine learning made the process easy and excellent.

This is an opensource project by NVIDIA.

Above you can see the sources A and B are combined to get some interesting images. Images generated are high resolution and scale specifically with a high level of details, thanks to machine learning.

source

Link

https://github.com/NVlabs/stylegan