Proposal for a PBL-type Learning Model Using 3D Gaussian Splatting

Abstract

This study proposes a next-generation problem-based learning (PBL) model for cultural tourism and education based on 3D Gaussian Splatting (3DGS), a technology that emerged in 2023. Compared with conventional photogrammetry and Neural Radiance Fields (NeRF), 3DGS enables high-definition and high-speed 3D rendering and real-time display on smartphones and stand-alone VR. In this study, this technology is not limited to a mere means of viewing, but is integrated with augmented intelligence (AI) to create a mechanism for dynamically presenting contextualized stories and learning materials to users. Specifically, using cultural resources in the Kirishima region of Kagoshima Prefecture as the subject matter, we propose a PBL model that integrates the entire process of filming, model optimization, guide presentation by AI, fieldwork, and sharing in the metaverse. In this process, the model was verified from various perspectives, including rendering delay, frame rate, learning effect, and ripple effect on the local economy. Students and tourists with prior virtual experience tended to significantly improve their learning depth and time spent at the site. Additionally, by converting the generated content to NFT and connecting it to local currencies and cultural heritage preservation funds, we propose a circular learning and economic ecosystem.