Cultural Heritage Anisotropic Objects (CHAO)

Overview

The Cultural Heritage Anisotropic Objects (CHAO) dataset is a synthetic benchmark dataset introduced in ShinyNeRF for evaluating neural rendering methods and inverse rendering algorithms on culturally significant objects with pronounced anisotropic reflectance. This dataset was created to support the research presented in ShinyNeRF, focusing on the challenges posed by complex geometry and environment lighting on heritage artifacts.

Dataset Description

CHAO contains renderings of two distinct cultural heritage objects: a Moroccan barad teapot and a German medieval helmet. Both objects exhibit strong anisotropic specularities characteristic of metallic craftsmanship. The 3D models were sourced from the Sketchfab catalog and processed to generate comprehensive ground truth data for material and geometry analysis under realistic lighting conditions.

Key Features

• Cultural heritage authenticity: Genuine artifact models provide realistic test cases for heritage digitization applications
• Environment map illumination: Complex HDR lighting captures realistic reflectance behavior under natural illumination conditions
• High-quality ground truth: Physically-based rendering provides accurate reference data for benchmarking on complex geometries

Dataset Contents

• RGB renderings: Rendered images (512×512 pixels) for each object under environment map lighting
• Material property maps: Ground truth maps including RGB, surface normals, tangent vectors, and depth (alpha)
• Blender scene files: Complete scene setups for reproducibility and custom rendering configurations

Objects

• Barad Teapot (Moroccan)
• Closed Helmet with Etched Decor (German medieval)

3D models source:All objects obtained from Sketchfab with modified material parameters (anisotropy and roughness). Model references: barad teapot, closed helmet