S2R-HDR: 24,000 Synthetic HDR Images for Reconstruction Tasks

24,000 high-resolution images across 1,000 motion sequences comprise this large-scale synthetic dataset for high dynamic range (HDR) reconstruction. S2R-HDR was rendered using Unreal Engine 5 and a custom pipeline built upon XRFeitoria. The dataset was last updated on March 2, 2026.

Use Cases

• Training HDR reconstruction models based on the 24-image motion sequences.
• Developing novel view synthesis algorithms based on the diverse dynamic scenes.
• Benchmarking image enhancement methods based on the linear HDR values stored in EXR format.
• Simulating real-world lighting conditions for computer vision tasks based on the Unreal Engine 5 rendering.

Strengths

• Contains 24,000 high-resolution images at 1920x1080 pixels.
• Includes 1,000 distinct motion sequences, each with 24 frames.
• Images are stored in EXR format with linear HDR values, supporting flexible data augmentation.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data is entirely synthetic, which may limit applicability to real-world scenarios.

Provenance

Source

Author iimmortall on Hugging Face.

Collection Method

Rendered using Unreal Engine 5 and a custom pipeline built upon XRFeitoria.

Time Range

null

Freshness

Last updated 2026-03-02 03:28:04; freshness should be verified.

Geography

null