Reflections Unlock: Geometry-Aware Reflection Disentanglement in 3D Gaussian Splatting for Photorealistic Scenes Rendering

2025-07-08

AI Summary

Conventional neural radiance fields (NeRF) and 3D Gaussian Splatting (3DGS) misinterpret reflective appearances as genuine geometric structure, leading to geometric inaccuracies in novel view synthesis. Method: We propose the first geometry-aware reflection-decoupling framework tailored for 3DGS. It employs a dual-branch radiance field to explicitly separate transmission and reflection components; incorporates pseudo non-reflection supervision, geometry-aware bilateral smoothing regularization, and high-order spherical harmonics for modeling high-frequency reflectance details; and integrates a reflection-removal module guided by pseudo depth maps to enhance reconstruction stability and fidelity. Contribution/Results: Our method significantly outperforms standard 3DGS on real-world reflective scenes, achieves NeRF-level rendering quality, and enables efficient, vision foundation model–driven reflection editing—while preserving geometric consistency and offering superior editability.

Technology Category

• Computer Vision: 3D Computer Vision
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Security and Privacy: Data transparency and provenance

Abstract

Accurately rendering scenes with reflective surfaces remains a significant challenge in novel view synthesis, as existing methods like Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS) often misinterpret reflections as physical geometry, resulting in degraded reconstructions. Previous methods rely on incomplete and non-generalizable geometric constraints, leading to misalignment between the positions of Gaussian splats and the actual scene geometry. When dealing with real-world scenes containing complex geometry, the accumulation of Gaussians further exacerbates surface artifacts and results in blurred reconstructions. To address these limitations, in this work, we propose Ref-Unlock, a novel geometry-aware reflection modeling framework based on 3D Gaussian Splatting, which explicitly disentangles transmitted and reflected components to better capture complex reflections and enhance geometric consistency in real-world scenes. Our approach employs a dual-branch representation with high-order spherical harmonics to capture high-frequency reflective details, alongside a reflection removal module providing pseudo reflection-free supervision to guide clean decomposition. Additionally, we incorporate pseudo-depth maps and a geometry-aware bilateral smoothness constraint to enhance 3D geometric consistency and stability in decomposition. Extensive experiments demonstrate that Ref-Unlock significantly outperforms classical GS-based reflection methods and achieves competitive results with NeRF-based models, while enabling flexible vision foundation models (VFMs) driven reflection editing. Our method thus offers an efficient and generalizable solution for realistic rendering of reflective scenes. Our code is available at https://ref-unlock.github.io/.

Problem

• Accurately render scenes with reflective surfaces
• Disentangle reflections from physical geometry
• Enhance geometric consistency in real-world scenes

Innovation

• Geometry-aware reflection modeling with 3D Gaussian Splatting
• Dual-branch representation using high-order spherical harmonics
• Pseudo-depth maps and geometry-aware smoothness constraint