The Use of Generative Adversarial Network as Graphical Support for Historical Urban Renovation

The history of small villages in inland areas inevitably also passes through the history of its renovation because it leads to new stimuli and drives design approaches. In this context, the large-scale planning and redevelopment of historic centers lead to demanding research and reflection work on the part of the planner. Recently, generative adversarial networks (GAN) have shown how they can be helpful for image generation following certain imposed rules, including those coming from architecture. This research work, exploiting the potentialities of Computational Geometry and automated information management through Data Engineering processes, aims to offer a graphical decision-support tool for urban renewal based on GAN to elaborate new color scales respecting the pre-existing ones. The case study in which the proposed approach is tested is the historic center of Caggiano Municipality, a small inland village in the Campania Region (Italy). The color scale generated was evaluated and compared with the color plan adopted at the time, with results very close to the traditional approach. Therefore, GANs represent valuable graphic data management support for designers who have to deal with historical contexts, especially in the initial study and context analysis phase.