NERF-DRIVEN ALGORITHMS APPLIED ON 360 IMAGES FOR 3D MODELLING OF HERITAGE ASSET IN VIRTUAL ENVIRONMENT

Abstract. This study aims to fill a gap in existing research by focusing on the application of Neural Radiance Fields (NeRF) algorithms to cultural heritage case studies using equirectangular image data captured with 360° cameras. The main objective is to evaluate and compare the performance of various NeRF algorithms applied to equirectangular images, shedding light on their suitability for cultural heritage preservation. the experiments were carried out on the case study of the library of the seminar in Brixen (Italy). By evaluating the effectiveness of NeRF in combination with conventional photogrammetric methods, the research highlights NeRF's competence in capturing complex details and addressing the challenges encountered in fast and expeditious 3D reconstruction of heritage. The positive results manifest in precise reconstructions, affirming the potential of NeRF in promoting the accuracy and fidelity of 3D models. Despite the computational demands, the study supports further exploration of NeRF-based algorithms, highlighting advantages and some limitations.