OpenSubstance: A High-Quality Measured Dataset of Multi-View and -Lighting Images and Shapes

Abstract

We present OpenSubstance, a high-quality measured dataset with 2.4 million high-dynamic-range images of 187 objects with a wide variety in shape and appearance, captured under 270 camera views and 1,637 lighting conditions, including 1,620 one-light-at-a-time, 8 environment, 8 linear and 1 full-on illumination. For each image, the corresponding lighting condition, camera parameters and foreground segmentation mask are provided. High-precision 3D geometry is also acquired for rigid objects. It takes 1 hour on average to capture one object with our custom-built high-performance lightstage and a top-grade commercial 3D scanner. We perform comprehensive quantitative evaluation on state-of-the-art techniques across different tasks, including single- and multi-view photometric stereo, as well as relighting.

Data Release

The dataset is divided into 10 parts and hosted on Science Data Bank. Part 1-10 all contain raw image folders, and Part 10 additionally includes:

Calibration data: camera parameters and light positions
Per-item data: capture metadata (ROI, lighting patterns), processed results (segmentation masks, COLMAP reconstructions), and aligned meshes (available for samples with high-quality scans)
Processing scripts: Python utilities for HDR merging, calibration data loading, and Blender format export

We recommend downloading the calibration data and scripts first, then selectively retrieving data for specific items as needed.

BibTeX

@InProceedings{Pei_2025_ICCV,
    author    = {Pei, Fan and Bai, Jinchen and Feng, Xiang and Bi, Zoubin and Zhou, Kun and Wu, Hongzhi},
    title     = {OpenSubstance: A High-quality Measured Dataset of Multi-View and -Lighting Images and Shapes},
    booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
    month     = {October},
    year      = {2025},
    pages     = {5221-5231}
}