| Property | Value |
|---|---|
| License | CreativeML OpenRAIL-M |
| Paper | arXiv:2405.14979 |
| Languages | English, Chinese |
| Pipeline Tag | Image-to-3D |
What is CraftsMan?
CraftsMan is a cutting-edge AI model designed for high-fidelity 3D mesh generation. Developed by researchers from HKUST and LightIllusions, it introduces a novel approach combining 3D native generation with an interactive geometry refiner. The model excels at converting 2D images into detailed 3D meshes while maintaining high geometric accuracy.
Implementation Details
The model utilizes a sophisticated pipeline that can be easily integrated into Python workflows using the CraftsManPipeline class. It supports both CUDA acceleration and runs on PyTorch with float32 precision. The implementation includes comprehensive features for mesh generation and export capabilities.
- Native PyTorch implementation with CUDA support
- Simple API for image-to-mesh conversion
- Interactive geometry refinement capabilities
- Support for standard 3D file formats (OBJ export)
Core Capabilities
- High-fidelity 3D mesh generation from 2D images
- Interactive geometry refinement for precise control
- Bilingual support (English and Chinese)
- Efficient processing with GPU acceleration
- Integration with popular 3D workflows
Frequently Asked Questions
Q: What makes this model unique?
CraftsMan stands out through its combination of native 3D generation and interactive geometry refinement, allowing for unprecedented control over the mesh generation process while maintaining high fidelity to the source material.
Q: What are the recommended use cases?
The model is particularly suited for applications requiring high-quality 3D mesh generation from 2D images, such as digital content creation, game development, and 3D visualization projects. It's especially valuable when precise geometric control is needed.