Make-Your-3D: Fast and Consistent Subject-Driven 3D Content Generation

Recent years have witnessed the strong power of 3D generation models, whichoffer a new level of creative flexibility by allowing users to guide the 3Dcontent generation process through a single image or natural language. However,it remains challenging for existing 3D generation methods to createsubject-driven 3D content across diverse prompts. In this paper, we introduce anovel 3D customization method, dubbed Make-Your-3D that can personalizehigh-fidelity and consistent 3D content from only a single image of a subjectwith text description within 5 minutes. Our key insight is to harmonize thedistributions of a multi-view diffusion model and an identity-specific 2Dgenerative model, aligning them with the distribution of the desired 3Dsubject. Specifically, we design a co-evolution framework to reduce thevariance of distributions, where each model undergoes a process of learningfrom the other through identity-aware optimization and subject-prioroptimization, respectively. Extensive experiments demonstrate that our methodcan produce high-quality, consistent, and subject-specific 3D content withtext-driven modifications that are unseen in subject image.