Unique Art Made by Artificial Intelligence

Explore

Featured Collections

About

What Makes My Art Special?

Most programmaticaly generated art that is being sold as NFT uses the same template over and over, each time randomly peacing predrawn components together. Imagine you got a set of 5 different eyes, lips, ears, and heads each. By random combinations 625 unique images can be created.
However, this concept lacks in originallity, as the components have to be compatible with each other and the results all have a certain similarity.

This is the point where the brilliance of Deep Neural Networks come into play: they are able to percieve images on a higher perceptual level and not simply as a set of pixels, as common programming does. These networks can learn deeper concepts and modify or generate content, shaping it to fit their learned concepts. This artistic freedom reflects in mindblowingly beatiful images that don't come with the 'copy and paste' look so many other NFTs are afflicted by.

How Does it Work?

Deep Neural Networks are very complex constructs, inspired by actual biological brains. If trained with enough data, these networks are capable of learning to understand how images are composed on a conceptual layer. Using special techniques, I train several networks to learn how famous painters used to paint, and let them apply their knowledge to conceptually change photos I have taken on my travels, transforming them into even more beautiful art.

How Unique Are Generated Images?

The knowledge of a Deep Neural Network is stored in its model weights. Normally, after the training is completed these weights are frozen and the network will always produce the same output for a given input. Being unique is what makes an NFT valuable in the first place, that's why I modified the generation process to require additional random inputs into specific layers of the network. This ensures a unique output even if the same input is used multiple times and even I, the creator, am not able to generate the same image twice.