NEURAL REPRESENTATION AND 3D DIGITAL SURVEY FOR INDUSTRIAL HERITAGE VALORISATION: THE CORRADINI FOUNDRY CASE STUDY (NAPLES)

The advanced digitisation of industrial heritage is a strategic lever for the regeneration of marginal urban areas, particularly where historical significance coexists with advanced degradation. This study applies and compares two 3D surveying and visualisation paradigms, photogrammetry and state-of-the-art neural methods (NeRF and 3D Gaussian Splatting), to the former Corradini foundry, a 5-hectare industrial-archaeological complex in San Giovanni a Teduccio (Naples), comprising 54 buildings that vary in age, function, and construction technique. The analysis indicates that NeRF/3DGS, owing to their capacity to deliver highly realistic spatial renderings even under unfavourable photometric conditions, can achieve greater perceptual fidelity than a traditional photogrammetric pipeline in this case study. These results open avenues for employing neural techniques in the interpretation, documentation and information management of stratified industrial contexts, supporting immersive modes of use and decision-making processes oriented toward sustainable reuse. The paper thus frames AI not merely as a technical instrument but as a cognitive interface between material memory and design.