Fantastic beasts and how to draw them with generative adversarial networks
Google revealed a demo of its Chimera Painter, a trained machine learning (ML) tool that automatically renders images of fantastic creatures using a user-generated outline.
Google positions the tool as an AI assistant allowing “game artists to quickly iterate on ideas and produce a high volume of assets, often in the face of tight deadlines.”
Here’s how it works. You produce a sketchy image of a creature with a brush, marking each body part with lables like “torso back” or “muzzle.” You can also import an image. Chimera Painter then adds textures and features using Google’s “CreatureGAN machine learning model, which was trained on hundreds of thousands of 2D renders of 3D creature models.”
Image Credit: Google
The trained GAN is now available in the Chimera Painter demo, allowing artists to work iteratively with the model, rather than drawing dozens of similar creatures from scratch. An artist can select a starting point and then adjust the shape, type, or placement of creature parts, enabling rapid exploration and for the creation of a large volume of images. The demo also allows for uploading a creature outline created in an external program, like Photoshop.
The tool is free. However, it’s still a prototype, so its real-life application is somewhat questionable. Below is the image of a sorry abomination Chimera Painter produced from my outline.
Nevertheless, the developers hope that the day will come when the tool can “reduce the amount of time necessary to create high-quality art without sacrificing artistic choices, perhaps even enhancing creativity.”
