Neural Piranesi is a series of artificially synthesised images exploring the work of the Italian architect and engraver Giovanni Battista Piranesi.

Famous for his etchings of Rome and of the fictitious ‘prisons’, Piranesi’s prints are renowned for revealing their subject matter in unprecedented and unimaginable ways. They could be original architectural creations or caprices of the imagination, or highly ornate and decorative models. He illustrated ancient relics and ruins, construction techniques and tools. The imaginative power of suggestion would permanently alter how people would emotionally perceive the ancient world.

Using the StyleGAN neural network, the complete works of Piranesi (just over 1000 works) were analysed and used to generate these AI etchings. This collection, the first in a series of three, demonstrates the trained network’s ability to create visually recognisable - though not replicant - elements of Piranesi’s style. Distinct subject typologies such as Piranesi’s vases, fireplaces, ruins and views can be clearly found, as can his stylistic etching textures, captions and annotations.

Every image is a high resolution (3.2K) jpg, each numbered in the lower right corner with a unique identifier which correlates to the images’ position in the latent space of the trained model.

About shedrawswithcode: Jessica In aka shedrawswithcode is an architect, designer, creative coder + educator based in London. Currently working freelance + in collaboration across architecture, installation + design projects. Teaching Lecturer at the Bartlett, where she is also a PhD candidate exploring the expressive potentials of machine learning for architectural drawing.