3D Models from Text Descriptions: Using Artificial Intelligence for Representation of Cultural Heritage
Francesca Condorelli · FrancoAngeli srl eBooks · 2025
This study explores the use of AI (Artificial Intelligence) to generate 3D models from textual descriptions and single-view image datasets, with the aim of digitally representing cultural heritage.The proposed approach combines advanced deep learning technologies, such as GAN and neural NeRF, to create 3D reconstructions from visual and textual data of limited quality and number.The use of textual descriptions overcomes the shortage of visual data and facilitates the reconstruction of architectural structures that are inaccessible because they no longer exist, have been destroyed or are only illustrated, offering a new way of digitally preserving architectural heritage.This innovative approach opens up new opportunities for the preservation and enhancement of cultural heritage, making it possible even in cases where historical structures are inaccessible.