Artistic GANs

In this lesson, we are going to explore the uses of GANs in the visual and sonic arts.

Visual arts

There are many GANs that produce impressive visual artifacts that can be used in the arts. Examples include the generation of paintings, anime characters, and fashionable clothes. Here, we provide a very small snippet of the visual work that is done using GANs.

GANGogh

GANGogh is the product of semester-long research performed by Kenny Jones and Derrick Bonafilia at Williams College in 2017.

In their project, the authors scoured the WikiArt database, which contains over 100,000 paintings, along with labels for style, genre, artist, and other features. The dataset that was used in their project contained 80,000 images with style and genre labels.

Their network setup is similar to the typical CGAN framework and is based on the improved WGAN. In the CGAN setup, the generator normally receives labels that are used to apply global conditioning on every layer; the discriminator, on the other hand, has an extra output that predicts the label of the input. In GANGogh, global conditioning was performed by applying a gated multiplicative activation function, similar to what is used in WaveNet and conditional PixelCNN.

Here are a few interesting image examples provided on their web page: