BlenderRAG: Procedural 3D Object Generation Scripts with Images and Descriptions

A multimodal dataset for 3D scene and object generation research. Each sample pairs a Blender Python script that procedurally generates a 3D object with a rendered preview image and a natural-language description. The dataset was created by MaxRondelli and last updated on Hugging Face in May 2026.

Use Cases

• Training models for text-to-3D generation based on natural language descriptions.
• Researching code-to-image synthesis based on Blender Python scripts.
• Developing multimodal retrieval systems based on aligned script, image, and text modalities.
• Benchmarking procedural generation techniques for indoor and outdoor scenes.

Strengths

• Dataset explicitly aligns three modalities per sample: Python code, rendered image, and text description.
• Organizes data into two top-level scene categories: indoor and outdoor.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

MaxRondelli on Hugging Face.

Collection Method

Likely procedurally generated using the Blender API.

Freshness

Last updated 2026-05-07 08:18:46; freshness should be verified.