The technology integrated with Artificial Intelligence functionality allows the acquisition of real spaces in an increasingly expeditious manner, giving the possibility of digitally configuring complex environments and experimenting with new digital surveying methodologies that differ from the established image/range-based techniques. In recent years, substantial advancements in the fields of computer graphics and computer vision have led to the emergence of innovative approaches, such as Gaussian Splatting, which have revolutionized 3D scene reconstruction and rendering processes. These methods offer remarkable improvements in both realism and computational efficiency. Moreover, in the digital age, the demand for accessible and user-friendly applications is constantly growing, and research is increasingly focused on solutions that offer essential functionality at cost-effective prices. These objectives are mainly pursued by exploiting open-source frameworks, cloud services, and simple methodologies to keep development costs low. The paper aims to validate the results of two Gaussian Splatting processes generated by accessible and user-friendly applications, as well as using simple data as a starting point, such as videos recorded with 360-degree cameras. In this way, the work seeks to evaluate the effectiveness of these innovative techniques in producing high-quality 3D reconstructions, considering the simplicity of the process for users and the associated costs.
Exploring Rapid 3D Heritage Asset Documentation: A Comparative Study of Laser Scanning and NeRF Algorithms in Museum Reconstruction
This paper explores the use of digital techniques for rapid surveys in cultural heritage, focusing on the creation of virtual environments for educational purposes. It compares two prominent methods for 3D reconstruction: laser scanning and AI-enhanced photogrammetry, particularly Neural Radiance Fields (NeRF). The case study centers on the virtual reconstruction of the Eccel Kreuzer Museum in Bolzano, South Tyrol, which involved documenting its original layout before an exhibition change. The research evaluates the performance of the Leica BLK laser scanner against the NeRF-based 3D models generated from video footage captured with an Insta360 ONE camera. While laser scanning provides high-quality results, the NeRF method, leveraging AI-based algorithms, offers a faster, cost-effective solution, particularly in environments with reflective surfaces and confined spaces. Despite challenges in accuracy and computational demands, the AI approach proves suitable for rapid documentation, especially when precision is less critical. The study highlights the advantages and limitations of both techniques, contributing to the ongoing development of 3D digital heritage assets that balance speed, cost, and quality in various applications such as VR and AR educational environments.
Rapid and Low-Cost 3D Model Creation Using Nerf for Heritage Videogames Environments
The paper explores the use of Neural Radiance Fields (NeRF) for the rapid and low-cost creation of 3D heritage environments intended for educational videogames. The research focuses on reconstructing architectural heritage using 360-degree images and videos processed through Nerfstudio and Nerfacto workflows. Compared with traditional photogrammetry and laser scanning, the proposed approach aims to reduce acquisition effort while maintaining visually convincing reconstructions. The study demonstrates how NeRF-based pipelines can generate point clouds and meshes suitable for immersive heritage education, while also discussing limitations related to computational requirements and reconstruction resolution.
Digital Turris Babel. Augmented Release of Athanasius Kircher’s Archontologia
This paper presents a digital heritage project focused on Athanasius Kircher’s Turris Babel (1679), aiming to reveal and communicate the hidden visual contents enclosed within the pages of the historical book. The research develops a workflow that combines iconographic analysis, image retrieval, customized photogrammetric reconstruction, and 3D modelling to recreate the imagined architecture of the Tower of Babel and other illustrated elements such as the labyrinth. Since the tower never existed physically and survives only through engraved representations, the reconstruction process addresses the challenge of modelling an architectural fiction from a single historical source. The resulting models are integrated into a markerless augmented reality application developed with Unity and Vuforia, allowing users to navigate and explore the structures directly from the printed pages. The study demonstrates how low-cost AR tools can enhance accessibility, interpretation, and public engagement with rare books and visual cultural heritage.
3DLAB SICILIA and UNESCO-VR. Models for Cultural Heritage
In the forecasts of 2019, the trend of the VR AR MR and AI sector was growing strongly; due to the pandemic events, real growth in the following two years has largely exceeded expectations. The possi-bility of interacting remotely by VR AR and MR modes during the pandemic was one of the solutions that made possible to overcome distances and barriers. As part of the 3DLAB Sicily project financed by the Sicily region and aimed at enhancing the resources of the cultural heritage of the territory, the De-partment of Civil Engineering and Architecture of the University of Catania (DICAR) participates in developing VR, AR, MR by a research team pertaining to the disciplines of design and representation, for the study of some UNESCO sites. The acquisitions of the models have been started using active and passive sensors for various cultural heritage; the case studies range from protohistoric archeology to medieval military architecture to the Baroque of the Val di Noto. The aim of the project is the creation of 3D navigable models for VR and AR viewers, or for two and three-walled VR caves that have been built as part of the same project. Below are the procedures, considerations and evaluations for the first results obtained.