RefRef: A Synthetic Dataset and Benchmark for Reconstructing Refractive and Reflective Objects

Yue Yin · Enze Tao · Weijian Deng · Dylan Campbell

Abstract

Modern 3D reconstruction and novel view synthesis approaches have demonstrated strong performance on scenes with opaque Lambertian objects. However, most assume straight light paths and therefore cannot properly handle refractive and reflective materials. Moreover, datasets specialized for these effects are limited, stymieing efforts to evaluate performance and develop suitable techniques. In this work, we introduce a synthetic RefRef dataset and benchmark for reconstructing scenes with refractive and reflective objects from posed images. Our dataset has 50 such objects of varying complexity, from single-material convex shapes to multi-material non-convex shapes, each placed in three different background types, resulting in 150 scenes. We also propose an oracle method that, given the object geometry and refractive indices, calculates accurate light paths for neural rendering, and an approach based on this that avoids these assumptions. We benchmark these against several state-of-the-art methods and show that all methods lag significantly behind the oracle, highlighting the challenges of the task and dataset.

Dataset Overview

RefRef is a synthetic dataset and benchmark designed for the task of reconstructing scenes with complex refractive and reflective objects. Our dataset consists of 50 objects categorized based on their geometric and material complexity: single-material convex objects, single-material non-convex objects, and multi-material non-convex objects, where the materials have different colors, opacities, and refractive indices. Each object is placed in three distinct bounded environments and one unbounded environment, resulting in 150 unique scenes with diverse geometries, material properties, and backgrounds. Our dataset provides a controlled setting for evaluating and developing 3D reconstruction and novel view synthesis methods that handle complex optical effects.

Directory Structure

RefRef/
├── data                          # parquet files
├── README.md
├── scene_list.txt                # A list of all 150 scenes
└── object_materials.json         # Optical material properties (IOR values) and categories for all objects

Object and Scenes

The dataset includes 50 objects categorised into three groups based on their complexity, material composition, and shape:

• single-convex(27 scenes): Objects with convex geometries, each composed of a single refractive material, such as transparent cubes, balls, and pyramids.
• single-non-convex(60 scenes): Objects with non-convex geometries, each composed of a single refractive material, such as animal sculptures, glass jars, light bulbs, candle holders, and magnifiers.
• multiple-non-convex(63 scenes): Objects with non-convex geometries, each composed of multiple refractive materials or a combination of refractive and opaque materials, such as reed diffusers, a glass of wine, and flasks filled with chemical liquid.

Each object is placed in three distinct scenes:

• textured_cube_scene: Objects placed within a bounded textured cube environment.
• textured_sphere_scene: Objects placed within a bounded textured sphere environment.
• environment_map_scene: Objects placed in an unbounded environment map background.

Citation

@misc{yin2025refrefsyntheticdatasetbenchmark,
      title={RefRef: A Synthetic Dataset and Benchmark for Reconstructing Refractive and Reflective Objects}, 
      author={Yue Yin and Enze Tao and Weijian Deng and Dylan Campbell},
      year={2025},
      eprint={2505.05848},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2505.05848}, 
}