Shap-E is a model by OpenAI that generates 3D assets from text prompts and images. The results can be rendered as textured meshes and neural radiance fields. When compared to similar models, Shap-E is much faster, but the generated models have lower detail.
| Advantages | Disadvantages |
|---|---|
| quick (13-60 seconds) | lower quality than competition |
| open source | problems when more than one object is created |
| can be combined with DreamFusion | complicated to set up for beginners (knowledge of git and Python notebooks necessary) |
| Blender knowledge highly recommended for exporting and improving the models |
To quickly create 3D models from images or a text prompt without the need for knowledge of modelling software. Only useful if detail is not important.
https://github.com/openai/shap-e
Prior knowledge of Python and Blender is highly recommended. Do not blindly follow this tutorial if you do not know what you are doing.