Digital Heritage Documentation: Integrating Photogrammetry, Gaussian Splatting, And 3D Mesh Modeling Practices for Enhanced Detail and Efficiency

Abstract

Nepal's rich architectural heritage is under increasing threat from environmental changes and rapid modernization, highlighting the urgent need for advanced digital preservation methods. Digital architectural and cultural heritage documentation has been evolving, yet the current practices remain limited to their ability to capture detailed structural representation and material nuances combined with its ability to scale and easily deploy. This study explores a hybrid approach that integrates photogrammetry, Gaussian splatting and contemporary 3D mesh modelling workflows to enhance visual fidelity and precise digital replicas. Photogrammetry is recognized for its geometric accuracy but can be computationally demanding and struggle to capture reflective and transparent surfaces. While Gaussian splatting offers fast and high- fidelity visualizations but it's integration with existing modelling pipelines is ongoing. By combining these methods, this study suggests a flexible and potentially more scalable workflow that leverages strength of each technique. Preliminary applications to small monuments indicate improvement in visual fidelity and processing efficiency that captures nuanced material details. Since this workflow requires only photos, it presents a low-barrier option for broader application. For the case study of this workflow, a set number of Nepalese architectural heritage and monuments will be photographed, processed and compared with existing workflows in terms of its visual fidelity \& computational demand.