Magic-Boost: Enhancing Coarse 3D Generation with Multi-View Conditioned Diffusion

Magic-Boost, a multi-view conditioned diffusion model, significantly refines coarse 3D generative results through a brief period of SDS optimization by leveraging precise guidance from synthesized multi-view images.