The advent of general adversarial networks (GANs) has led to increased popularity and adoption of artificial intelligence in the art world. It has been quite a few years since researchers have been trying to infuse the artistic skills into AI and there have been many interesting developments since then. Artists such as Mario Klingemann, Anna Ridler and many others have been at the forefront of this new-age GAN-powered art. Not only is AI creating breathtaking artwork but it is also being sold at auctions for hefty amounts. For instance, Canadian-Mexican artist Rafael Lozano-Hemmer has already made around $600,000 for an AI artwork.
In this article, we are listing down some of the top AI-based artworks that made headlines in 2019:
Sign up for your weekly dose of what's up in emerging technology.
1| AI Paintings By Mauro Martino
An algorithm created by researcher Mauro Martino can create portraits entirely from scratch. Instead of the usual style transfer, it uses the generative adversarial network to conjure up new, Renaissance-like paintings that evoke human likeness. An artist-in-residence at the MIT-IBM Watson AI Lab and a professor at Northeastern University, his team has trained GAN algorithm on 45,000 portraits in the traditional style of Renaissance portraiture. It can not take inspiration from previous paintings and create something new, but can also be directly inspired by looking at the picture of a person. The website allows to upload a picture and it generates a unique work of art painted by an experienced human painter. It currently deletes the photos immediately from the project’s server to protect user’s privacy.
2| AI Artwork By NVIDIA
NVIDIA Research has also explored its expertise in AI-based arts as a deep learning model developed by the company can turn rough doodles into photorealistic masterpieces with breathtaking ease. This tool also leverages GANs to convert segmentation maps into lifelike images. Called GauGAN, this interactive app allows the user to draw simple sketches and doodles while labelling each segment with labels like sand, sky, sea or snow which is then processed into life-like images by GANs. Developed using the PyTorch deep learning framework, it has been gaining a lot of popularity in the community.
3| AI Robot That Can Paint
A different twist to the AI-based painting, this interesting innovation is all about the world’s first ultra-realistic robot artist that can draw and paint using pen and pencil and creates sketches in a human-like manner. Developed by Aidan Meller, the first exhibition was displayed at the University of Oxford, in July 2019. Named Ai-Da after Ada Lovelace, the first female computer programmer in the world, she uses facial recognition technology to draw human faces. With her own personal voice, she calls Picasso’s Guernica and Orwell’s 1984 as key inspirations for her artistic practice. Invented by Meller and built by robotics company Engineered Arts, she is infused with AI programming from PhD students at Oxford University.
4| AI Paintings By Chris Peters
Another artist Chris Peters created his own AI Muse and used it to create paintings of landscapes that do not actually exist. Peters fed the system with a lot of landscape paintings curated from different sources, using which the algorithms learnt and predicted, and created new digital images after processing these landscapes. Unveiled in March 2019, at a private exhibition, these AI-based paintings gained major traction and shows an innovative collaboration between man and machine.
5| AI Paintings By Barrat And Barrot
In another interesting development, artist Robbie Barrat collaborated with a painter named Ronan Barrot who have been training GAN on the latter’s paintings of skulls. The output produced by GANs were displayed alongside the original paintings. The show titled “BARRAT/BARROT: Infinite Skulls” hosted some interesting work by the two artists and GANs. As Robbie shares, he trained the network on the skulls. They are all the same shape, size and orientation. The results were very similar to Ronan’s original skulls.
6| Mona Lisa Was Brought To Life
AI researchers at Samsung brought the epic Mona Lisa by Leonardo Da Vinci to life to developing a way to animate classic portrait. The researchers from Samsung’s AI centre in Moscow, along with the Skolkovo Institute of Science and Technology used deepfake techniques which is ideally used to graft one person’s face onto a video of someone else. The software was fed with a huge dataset of celebrity talking-head videos from YouTube and pinpointing significant landmarks on human faces, for which GANs were used.
7| AI Painting From Text Commands
In another development, AI was seen to be producing artwork just by following a person’s text-based command. When instructions are passed in plain English, the model comes up with an understandable art with a boy, girl, sandbox, hot dog and a shovel. Developed by researchers at the University of Virginia in collaboration with IBM Watson, Text2Scene, as the model is called, generates various forms of compositional scene representations from natural language descriptions. Researchers also claim that Text2Scene model is competitive and also can produce superior interpretable results.
8| Memories of Passersby I By Mario Klingemann
This art piece created completely by AI was put up at Sotheby’s Contemporary Art Day Auction in March 2019. Completely a machine installation, it uses neural networks to generate an infinite stream of portraits. It uses a complex system of neural networks to generate unique uncanny representations of male and female faces. Each portrait is created in real-time as machine interprets its own output. Klingemann’s AI was trained using thousands of portraits from the 17th to 19th centuries.