Word pictures: new insights through AI around the villa Laurentina by Pliny the Younger

Abstract

Several renowned architectural masterpieces, preserved solely through literary descriptions, have played a crucial role in facilitating discussions on both theoretical and visual perspectives, offering insights about their real appearance. Regardless of the level of detail in textual descriptions, words fall short in fully encapsulating the multifaceted nature of architecture. This inherent lack of precise definition opens up a creative space for authors who embark on the challenge of visually translating text into images. In doing so, theoreticians and architects have infused their creativity into the graphic representation of those descriptions, enriching the visual interpretation of architectural concepts. The task of converting textual descriptions into images is now closely aligned with the new capabilities of text-to-image generation provided by numerous Artificial Intelligence applications. This study aims to conduct a comprehensive review and update of the Villa Laurentina through experiments with “prompts” derived from the texts of Pliny the Younger, in order to generate and evaluate the resulting images. Those capabilities enable us to suggest as well an exercise in observing how Stable Diffusion reacts once it has been fine-tuned with graphic proposals that have been issued by scholars throughout history, incorporating those same “prompts” by Pliny. The outcomes are analyzed in terms of the ability of artificial intelligence to decode texts and perform visual interpretations about the classical theme of representing lost architectures after customizing the Stable Diffusion tool.