Real-time rendering and interaction of 3D Gaussian splatting scenes in Unreal Engine for mixed reality

To enhance interaction understanding in mixed reality (MR) environments, this study proposes an MR interaction and measurement system based on three-dimensional Gaussian splatting (3DGS). The system leverages the efficient spatial representation capability of 3DGS to achieve real-time rendering in Unreal Engine (UE), and supports precise geometric measurement by converting 3DGS data into dense point clouds. Experimental results on the Mip-NeRF360 and Tanks and Temples datasets demonstrate that the system can achieve real-time rendering at over 60 FPS while preserving scene realism, and also supports instant area measurement. These results validate the feasibility of integrating 3DGS with dense point clouds for MR interaction and measurement, providing a reliable data foundation for downstream applications such as digital twins, virtual surveying, and architectural simulation.