Magic123: One Image to High-Quality 3D Object Generation Using Both 2D and 3D Diffusion Priors

TLDR

Magic123 is a two‑stage method that generates high‑quality, textured 3D meshes from a single unposed image by leveraging both 2D and 3D diffusion priors. The method first optimizes a neural radiance field for coarse geometry, then refines it with a memory‑efficient differentiable mesh representation, guided by reference view supervision, 2D and 3D diffusion priors, a trade‑off parameter, textual inversion, and monocular depth regularization to produce high‑resolution textured meshes. Magic123 significantly outperforms prior image‑to‑3D techniques on synthetic benchmarks and diverse real‑world images. Code, models, and generated 3D assets are available at https://github.com/guochengqian/Magic123.

Abstract

We present Magic123, a two-stage coarse-to-fine approach for high-quality, textured 3D meshes generation from a single unposed image in the wild using both2D and 3D priors. In the first stage, we optimize a neural radiance field to produce a coarse geometry. In the second stage, we adopt a memory-efficient differentiable mesh representation to yield a high-resolution mesh with a visually appealing texture. In both stages, the 3D content is learned through reference view supervision and novel views guided by a combination of 2D and 3D diffusion priors. We introduce a single trade-off parameter between the 2D and 3D priors to control exploration (more imaginative) and exploitation (more precise) of the generated geometry. Additionally, we employ textual inversion and monocular depth regularization to encourage consistent appearances across views and to prevent degenerate solutions, respectively. Magic123 demonstrates a significant improvement over previous image-to-3D techniques, as validated through extensive experiments on synthetic benchmarks and diverse real-world images. Our code, models, and generated 3D assets are available at https://github.com/guochengqian/Magic123.