cvlab-columbia/zero123
Zero-1-to-3 is a zero-shot model that converts a single 2D image into a 3D object representation, enabling novel view synthesis and 3D reconstruction.
Not currently ranked — collecting fresh signals.
star history
This research project from Columbia University and Toyota Research Institute leverages stable diffusion to perform zero-shot 3D reconstruction from a single image. The model generates multiple views of an object from different camera angles, enabling downstream 3D asset creation without per-category training. It extends 2D diffusion models to 3D understanding tasks by conditioning on relative camera viewpoint information.
Frequently asked
- What is cvlab-columbia/zero123?
- Zero-1-to-3 is a zero-shot model that converts a single 2D image into a 3D object representation, enabling novel view synthesis and 3D reconstruction.
- Is zero123 open source?
- Yes — cvlab-columbia/zero123 is open source, released under the MIT license.
- What language is zero123 written in?
- cvlab-columbia/zero123 is primarily written in Python.
- How popular is zero123?
- cvlab-columbia/zero123 has 3.1k stars on GitHub.
- Where can I find zero123?
- cvlab-columbia/zero123 is on GitHub at https://github.com/cvlab-columbia/zero123.