Latent Places is a generative machine learning art project consisting of 60 compositions created with CLIP + VQGAN. Each image shows an AI-composed scene inspired by our natural world.

Each piece is minted as a unique 1/1 NFT.

The CLIP + VQGAN technique is an AI art technique in which the artist describes an image in natural language to a system of machine learning models. The machine learning system attempts to create an image matching the natural language prompt based on a semantic understanding of natural language and image composition learned by the models during an initial training process. The resulting image is a synthesis between the artist's ability to compose a compelling text prompt and the machine's understanding of the human world.