A new AI tool called Wonder3D allows users to generate highly detailed, textured 3D meshes from a single 2D image.
The technology uses a cross-domain diffusion model. First, it generates consistent multi-view normal maps and corresponding color images from the input. Then, it employs a novel normal fusion method to rapidly produce a high-quality 3D reconstruction. This workflow is particularly valuable for artists and designers: instead of manually modeling objects from multiple angles, they can start with one photo and get a usable 3D asset in just a few minutes. Wonder3D can be integrated into existing creative pipelines, such as those using ComfyUI.
About the Author
