GaussianShopVR: Facilitating Immersive 3D Authoring Using Gaussian Splatting in VR

Yulin Shen1, Boyu Li1, Jiayang Huang1, David Yip1, Zeyu Wang1,2
The Hong Kong University of Science and Technology (Guangzhou)1, The Hong Kong University of Science and Technology2

Abstract

We present GaussianShopVR, a VR-based authoring system for controllable and fine-grained editing of 3D Gaussian Splatting (3DGS). Although 3DGS has gained popularity for its capability to quickly create digital replicas of real-life scenes and its compatibility with existing rendering pipelines, current methods often struggle with detailed 3D editing due to cumbersome target area selection and inadequate spatial control on AI-driven editing. GaussianShopVR addresses these challenges by harnessing intuitive VR interactions for efficient point selection and drawing. We further enhance editing methods to support interactive creative tasks, including precise object splitting, real-time color adjustment, and controllable object generation from drawn points. We evaluate GaussianShopVR through three user studies focusing on point selection efficiency (N=18), controllable 3D generation (N=20), and overall usability (N=10). The findings suggest that GaussianShopVR provides an immersive, flexible, and controllable approach to 3DGS-based content creation in VR. Our code is available at https://github.com/CISLab-HKUST/GaussianShopVR.

PDF BibTeX
BibTeX copied to clipboard