search

cvlab-columbia / zero123

Zero-1-to-3: Zero-shot One Image to 3D Object (ICCV 2023)
https://zero123.cs.columbia.edu/
MIT License
2.58k stars 188 forks source link

Can model trained to generate human #94

Open Dragonkingpan opened 9 months ago

Dragonkingpan commented 9 months ago

A very excellent work. I want to use this technology to generate a 3D model of the human body from a single photo. Because we already have code for binding bones to human 3D models and automatically brushing weights, and the results are quite good, especially for people with A Pos. If a frontal photo is used, a digital person with good accuracy can be generated and can automatically move. Is this a very interesting thing. I think for something with a very obvious and fixed type of human characteristic, the training results should

BaochaoZhu commented 3 months ago

@Dragonkingpan ,Do you solve it? I have same problem?