LightCity
:

An Urban Dataset for Outdoor Inverse Rendering and Reconstruction under Multi-illumination Conditions

1State Key Lab of CAD & CG, Zhejiang University
Corresponding Author StarEqual Contribution     Corresponding Author Star Corresponding Author
ICCV 2025
Paper Code Dataset
overview
LightCity is a novel high-quality synthetic urban dataset. It features complicated urban illumination conditions, including varied illumination, realistic indirect lighting and shadow effects, and varying scales with street and aerial image capture.

Abstract

Inverse rendering in urban scenes is pivotal for applications like autonomous driving and digital twins. Yet, it faces significant challenges due to complex illumination conditions, including multi-illumination, indirect light and shadow effects. However, the effects of these challenges on intrinsic decomposition and 3D reconstruction have not been explored due to the lack of appropriate datasets. We present LightCity, a novel high-quality synthetic urban dataset featuring diverse illumination conditions with realistic indirect light and shadow effects. LightCity encompasses over 300 sky maps with highly controllable illumination, varying scales with street-level and aerial perspectives over 50K images, and rich properties such as depth, normal, material components, light and indirect light, etc.

Overview

    LightCity has the following key features:
  • High quality: LightCity is built upon Blender Cycle engine to deliver photo-realistic images with realistic lighting and shadow effects.
  • Rich illumination diversity: We incorporate over 300 sky maps spanning the entire day, from dawn to night, providing high controllability in illumination through adjustable rotation and intensity of HDRI maps.
  • Scale: The dataset encompasses synthetic urban blocks of varying scales, with over 30K views for intrinsic tasks and over 20K views for reconstruction tasks, covering both street-level and aerial perspectives.
  • Comprehensive Properties: The dataset includes multiple attributes that can support various vision tasks, such as depth and normal maps for geometry estimation, diffuse and glossy components for material estimation, etc.
pipeline

Fig. 1: Dataset Overview. The features of our urban datasets: (a) Diverse variety and flexible control of illuminations. (b) View sampling with varying scales. (c) Multiple properties.

Color
Intensity
Shading

Benchmark

We benchmark on three foundamental tasks: intrinsic image decomposition, multi-view inverse rendering and multi-illumination urban reconstruction.

• Multi-illumination Reconstruction

benchmark reconstruction

Tab. 1: Comparisons of novel view synthesis under multi-illumination in urban scenes.

• Intrinsic Image decomposition

benchmark reconstruction

• Multi-view Inverse Rendering

benchmark inverse rendering
inverse rendering quality

BibTeX

@inproceedings{wang2025lightcity,
  title={LightCity: An Urban Dataset for Outdoor Inverse Rendering and Reconstruction under Multi-illumination Conditions},
  author={Jingjing Wang, Qirui Hu, Chong Bao, Yuke Zhu, Hujun Bao, Zhaopeng Cui, Guofeng Zhang},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
  year={2025},
}