3D Generative

3D generative modeling aims to create realistic three-dimensional objects and scenes from various inputs like text, images, or point clouds. Current research heavily utilizes diffusion models, often coupled with efficient 3D representations such as Gaussian splatting or signed distance functions, to generate high-fidelity outputs with improved speed and control. This field is significant for its potential to automate 3D content creation across diverse applications, from virtual reality and gaming to robotics and drug discovery, while also advancing our understanding of generative AI itself.