Today we are checking out Artbreeder, a machine learning (GAN – Generational Adversarial Network) based art creation tool. Using Artbreeder you can rapidly create portraits, landscapes, characters and more using sampled (or uploaded) existing art, a series of “genes” as well as crossbreeding with other source images or art styles. The results are truly astonishing and this could be an excellent tool for generation game portraits, concept art and more.
Artbreeder is described as:
Artbreeder aims to be a new type of creative tool that empowers users creativity by making it easier to collaborate and explore. Originally Ganbreeder, it started as an experiment in using breeding and collaboration as methods of exploring high complexity spaces. Artbreeder is named after the research of Picbreeder which investigated the role of exploration in the optimization process. It is also inspired by an earlier project of mine Facebook Graffiti which demonstrated the creative capacity of crowds.
There is a minimalistic open source implementation of the logic powering Artbreeder available here on GitHub under the GPL v3 license, in case you wanted to peek under the hood to see how it worked. If you are interested in using Artbreeder in a commercial project, their are several commercial tiers available:
You can see this amazing tool in action in the video below.