DreamCraft3D++: Efficient Hierarchical 3D Generation with Multi-Plane Reconstruction Model
We introduce DreamCraft3D++, an extension of DreamCraft3D that enables efficient high-quality generation of complex 3D assets. DreamCraft3D++ inherits the multi-stage generation process of DreamCraft3D, but replaces the time-consuming geometry sculpting optimization with a feed-forward multi-plane based reconstruction model, speeding up the process by 1000x. For texture refinement, we propose a training-free IP-Adapter module that is conditioned on the enhanced multi-view images to enhance texture and geometry consistency, providing a 4x faster alternative to DreamCraft3D's DreamBooth fine-tuning. Experiments on diverse datasets demonstrate DreamCraft3D++'s ability to generate creative 3D assets with intricate geometry and realistic 360° textures, outperforming state-of-the-art image-to-3D methods in quality and speed. The full implementation will be open-sourced to enable new possibilities in 3D content creation.
Authors
- Ying Guo (ORCID: https://orcid.org/0000-0001-7651-6087)
- Jie Zhang (ORCID: https://orcid.org/0000-0001-7821-4808)
- Yangguang Li (ORCID: https://orcid.org/0000-0002-9696-8192)
- Yebin Liu (ORCID: https://orcid.org/0000-0003-3215-0225)
- Cheng Peng (ORCID: https://orcid.org/0000-0002-4605-7534)
- Xiaochen Zhao (ORCID: https://orcid.org/0000-0002-6603-257X)
- Ruizhi Shao (ORCID: https://orcid.org/0000-0003-2188-1348)
- Yan‐Pei Cao (ORCID: https://orcid.org/0000-0002-0416-4374)
- Jingxiang Sun (ORCID: https://orcid.org/0009-0003-9916-9585)
Institutions
- Zhejiang University (CN)
Publication Details
- Journal
- IEEE Transactions on Pattern Analysis and Machine Intelligence
- Published
- 2025-01-01
- DOI
- https://doi.org/10.1109/tpami.2025.3598772
- Citations
- 1
- Primary Topic
- Computer Graphics and Visualization Techniques
- Type
- article
- Field-Weighted Citation Impact
- 3.54