New videos from the EuroPython 2019 conference were recently uploaded to YouTube. There are currently 133 videos with a wide range of Pythonic topics. The keynote on July 11, 2019, “AI in Contemporary Art” was delivered by Luba Elliott (@elluba). Elliott’s informative, entertaining and engaging delivery of a beautifully curated set of AI artworks is a must-see. Here are a few topics from the talk (with lots of links!):
- Deep Dream – The talk starts with Deep Dream generators and Memo Atken. If you haven’t played with Deep Dream yet, here are a few recent posts to get up to speed. The images are surreal psychedelic fun (lots of eyes and slugs).
- Style Transfer – Elliott hits on this method showing landscape images converted into Picaso, Monet or Contemporary art style images. If you’d like to play with some generative adversarial networks (GAN) and style transfer, GauGAN has a number of fun options. Gene Kogan is mentioned as taking the method of style transfer a bit further creating works like the Mona Lisa in the style of Google Maps.
- GANs – GANs utilize two opposing models to create highly realistic results (images, translations, etc). The artist Mario Klingemann is mentioned here as being prolific in the use of different GANs. Elliott muses that the Klingemann’s GAN images are [surreal] and remind the viewer of Francis Bacon.
- Data Set Manipulation – Other artists have experimented with curating the data sets for models. In Roman Lipski’s case, images of LA at night. Lipski created several paintings of the scene, a model was trained on those paintings, and Lipski used the model’s results as inspiration for new paintings. This went on in a feedback loop to create the project ‘unfinished‘. Elliott comments that Lipski, “developed stylistically through the help of the algorithm”.
- Sculpture – Ben Snell utilized machine learning to create plans for the sculpture, “Dio“. Snell then used the computer (also named Dio) as the material for the piece. “However, this sculpture possesses a secret: it is not only made by the computer, but materially made from the computer that dreamt it.”
- Facial Recognition – This controversial technology has stirred a number of projects aimed at confusing or spoofing the AI. Constant Dullart created “DullDream” which dulls or reduces the features in an image (take a look at the photos above).
- AI Art Robot – Last but not least, AIda the AI art Robot! AIda is a robot that can paint and sketch.
If you’d like to learn more about AI in art checkout Elliott’s website. If you are interested in attending an AI art workshop, checkout NeuroIPS: “The goal of this workshop is to bring together researchers interested in advancing art and music generation to present new work, foster collaborations and build networks.” The workshop will be held in Vancouver, Canada on Saturday, December 14th from 8:30 to 18:00.