Solving light parameters using RadiosityGS | TU Wien – Research Unit of Computer Graphics

Information

Publication Type: Student Project
Workgroup(s)/Project(s):
- Palaces to Pixels: Advanced 3D Reconstruction for Heritage
Date: 2026
Date (Start): October 2025
Date (End): 3. September 2026
Matrikelnummer: 12507307
First Supervisor: Philipp Erler
Keywords: Light-source estimation, Gaussian Splatting, RadiosityGS

Abstract

This student project investigates light-source estimation in 3D using only photos as inputs. It can accurately optimize the light properties even when the lights are occluded.

Additional Files and Images

Additional images and videos

image: Optimization result for one light source on the hotdog scene

Additional files

Report

Weblinks

Code

BibTeX

@studentproject{pakkanen-2026-LightEst,
  title =      "Solving light parameters using RadiosityGS",
  author =     "Eetu Pakkanen",
  year =       "2026",
  abstract =   "This student project investigates light-source estimation in
               3D using only photos as inputs. It can accurately optimize
               the light properties even when the lights are occluded.",
  month =      sep,
  keywords =   "Light-source estimation, Gaussian Splatting, RadiosityGS",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2026/pakkanen-2026-LightEst/",
}