PanoX: Large-Scale Synthetic Panorama Dataset for 3D Scene Generation

PanoX is a large-scale synthetic panorama dataset constructed to advance 2D lifting techniques for generating graphics-ready 3D scenes. It features high-quality multimodal panoramas from diverse indoor environments, supporting physically based rendering, relighting, and simulation. The dataset was created by KevinHuang for the paper "OmniX: From Unified Panoramic Generation and Perception to Graphics-Ready 3D Scenes" and was last updated in March 2026.

Use Cases

• Training models for 2D-to-3D scene lifting based on the described high-quality multimodal panoramas.
• Developing physically based rendering (PBR) pipelines based on the graphics-ready 3D scene data.
• Researching scene relighting and simulation techniques based on the synthetic indoor panorama environment.
• Benchmarking panoramic perception and generation models based on the large-scale, diverse dataset.

Strengths

• Dataset is described as large-scale, suggesting substantial volume.
• Features high-quality multimodal panoramas from diverse indoor environments.
• Explicitly designed for graphics-ready 3D scenes suitable for PBR, relighting, and simulation.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and exact size are unknown, which may limit suitability assessment.
• Data is synthetic, which may not fully capture the complexity of real-world scenes.

Provenance

Source

KevinHuang on Hugging Face, associated with the paper "OmniX: From Unified Panoramic Generation and Perception to Graphics-Ready 3D Scenes".

Collection Method

Synthetically constructed, as described.

Freshness

Last updated 2026-03-30 06:38:56; freshness should be verified.