HSI Nerfstudio Data: DIRSIG-LWIR-HSI-SF6 Plume (multi-angle dataset of simulated gas plume hyperspectral imagery)

HSI Nerfstudio Data: DIRSIG-LWIR-HSI-SF6 Plume This dataset contains DIRSIG-simulated longwave infrared (LWIR) hyperspectral imagery of a simplified facility scene designed for testing 3D scene understanding and NeRF-based reconstruction methods using hyperspectral data. The scene was created by the DIRSIG team at Rochester Institute of Technology (RIT). It includes a smokestack, buildings, and surrounding infrastructure. A sulfur hexafluoride (SF₆) gas plume emanates from the smokestack. The dataset is intentionally simple and intended as a controlled test environment for hyperspectral 3D reconstruction and gas plume analysis. To see examples of the images, preview the `False_Color_Examples.jpg` and `Ground_Truth_Example.jpg` files, or go to https://github.com/lanl/HSI-Nerfstudio. Dataset Specifications 231 hyperspectral images Spatial resolution: 128 × 128 pixels Spectral resolution: 128 channels Spectral range: 7–13 µm (MAKO sensor model) Associated ground truth data for geometry and plume properties Camera extrinsics and intrinsics metadata Reference absorption spectrum for SF₆ Dataset Contents 1. Spectral Reference sf6_ext.txt: Library reference absorption spectrum of sulfur hexafluoride (SF₆), including wavenumber and absorption values. 2. Image Data ENVI-formatted hyperspectral images (.img and .hdr) Corresponding ground truth files (truth-*.img and .hdr) Compatible with standard ENVI readers (e.g., spectral Python package) 3. Camera Metadata transforms.json containing camera intrinsics and extrinsics formatted for use with hsi-nerfstudio 4. Camera Paths JSON files defining camera trajectories for NeRF rendering Intended Use This dataset was used in SPIE publications investigating hyperspectral NeRF reconstruction and gas plume modeling. It is intended for research in: Hyperspectral imaging Neural Radiance Fields (NeRFs) 3D scene reconstruction Spectral gas detection Remote sensing simulation validation Associated codebase: hsi-nerfstudio (GitHub). For use with hsi-nerfstuio, simple download and unzip the data.zip file into your repository. LA-UR-25-30099