Skyfall-GS: 3D Urban Scene Datasets for JAX and NYC

JAX and NYC datasets for training and evaluating Skyfall-GS, a hybrid framework for synthesizing city-block scale 3D urban scenes. The data was uploaded by author jayinnn and last updated on March 18, 2026. The datasets combine satellite reconstruction with diffusion refinement.

Use Cases

• Training 3D scene synthesis models based on satellite imagery and diffusion refinement.
• Evaluating the quality of generated immersive urban environments.
• Benchmarking hybrid reconstruction frameworks that combine different data sources.

Strengths

• Dataset is associated with a published research paper and project page, suggesting academic rigor.
• Data is organized into specific geographic subsets (JAX and NYC) for targeted evaluation.

Limitations

• Dataset size, row count, file formats, and column structure are unknown.
• Column-level documentation is absent; field semantics must be inferred after download.
• License information is unknown, which may restrict usage.

Provenance

Source

jayinnn on Hugging Face, associated with the Skyfall-GS project.

Collection Method

Likely contains data synthesized by combining satellite reconstruction with diffusion refinement.

Freshness

Last updated 2026-03-18 10:11:48; freshness should be verified.

Geography

Covers at least two locations: JAX and NYC.