Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

GLINT: Modeling Scene-Scale Transparency via Gaussian Radiance Transport

Youngju Na, Jaeseong Yun, Soohyun Ryu, Hyunsu Kim, Sung-Eui Yoon, Suyong Yeon

Abstract

While 3D Gaussian splatting has emerged as a powerful paradigm, it fundamentally fails to model transparency such as glass panels. The core challenge lies in decoupling the intertwined radiance contributions from transparent interfaces and the transmitted geometry observed through the glass. We present GLINT, a framework that models scene-scale transparency through explicit decomposed Gaussian representation. GLINT reconstructs the primary interface and models reflected and transmitted radiance separately, enabling consistent radiance transport. During optimization, GLINT bootstraps transparency localization from geometry-separation cues induced by the decomposition, together with geometry and material priors from a pre-trained video relighting model. Extensive experiments demonstrate consistent improvements over prior methods for reconstructing complex transparent scenes.

GLINT: Modeling Scene-Scale Transparency via Gaussian Radiance Transport

Abstract

While 3D Gaussian splatting has emerged as a powerful paradigm, it fundamentally fails to model transparency such as glass panels. The core challenge lies in decoupling the intertwined radiance contributions from transparent interfaces and the transmitted geometry observed through the glass. We present GLINT, a framework that models scene-scale transparency through explicit decomposed Gaussian representation. GLINT reconstructs the primary interface and models reflected and transmitted radiance separately, enabling consistent radiance transport. During optimization, GLINT bootstraps transparency localization from geometry-separation cues induced by the decomposition, together with geometry and material priors from a pre-trained video relighting model. Extensive experiments demonstrate consistent improvements over prior methods for reconstructing complex transparent scenes.

Paper Structure

This paper contains 47 sections, 17 equations, 16 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Gaussian Splatting
  4. Radiance Decomposition in Gaussian Splatting
  5. Transparent Scene Reconstruction
  6. Method
  7. Preliminaries: 2D Gaussian Rendering
  8. Rendering techniques.
  9. Decomposed Gaussian Representation
  10. Hybrid rendering pipeline.
  11. Transparency-Aware Radiance Transport
  12. Opaque branch ($L_{\text{opaque}}$).
  13. Transparent branch ($L_{\text{transparent}}$).
  14. Optimization
  15. Geometric regularization.
  16. ...and 32 more sections

Figures (16)

  • Figure 1: Pipeline Overview. The interface component $\mathcal{G}_{\text{intr}}$ captures primary first-surface, while secondary components $\mathcal{G}_{\text{trans}}$ and $\mathcal{G}_{\text{refl}}$ separately model transmission and reflection. The output color $L_o$ is obtained through hybrid rendering under transparency-aware radiance transport and supervised by photometric loss $\mathcal{L}_{\text{photo}}$. DiffusionRenderer liang2025diffusion provides priors that regularize the G-buffers via $\mathcal{L}_{\text{geo}}$.
  • Figure 2: Depth decomposition. (Left) Rendered image, (Middle) interface depth, and (Right) transmission depth.
  • Figure 3: Obtained transparency maps. GT images (first row) and learned transparency maps (second row).
  • Figure 4: Qualitative comparison on synthetic scenes. Each column shows results from GT, PGSR chen2024pgsr, EnvGS xie2025envgs, TSGS li2025tsgs, and Ours. For each scene, rows correspond to RGB (top) and depth maps (bottom).
  • Figure 5: Mesh visualization comparison. The meshes are obtained from TSDF fusion following baselines.
  • ...and 11 more figures