OctFusion: Octree‐based Diffusion Models for 3D Shape Generation
Computer Graphics Forum · 2025-08 · 44 (5)
DOI: 10.1111/cgf.70198 ↗Kurz gefasst
Fachartikel in der Zeitschrift Computer Graphics Forum, online erschienen am 28.08.2025. Verfasst von Bojun Xiong und 5 weiteren Autor:innen. Die Arbeit wurde bisher 22-mal zitiert (Crossref) und nennt 73 Literaturangaben. Gefördert von: National Key Research and Development Program of China, Natural Science Foundation of Beijing Municipality, National Natural Science Foundation of China.
Abstract (Originalsprache)
Diffusion models have emerged as a popular method for 3D generation. However, it is still challenging for diffusion models to efficiently generate diverse and high‐quality 3D shapes. In this paper, we introduce OctFusion, which can generate 3D shapes with arbitrary resolutions in 2.5 seconds on a single Nvidia 4090 GPU, and the extracted meshes are guaranteed to be continuous and manifold. The key components of OctFusion are the octree‐based latent representation and the accompanying diffusion models. The representation combines the benefits of both implicit neural representations and explicit spatial octrees and is learned with an octree‐based variational autoencoder. The proposed diffusion model is a unified multi‐scale U‐Net that enables weights and computation sharing across different octree levels and avoids the complexity of widely used cascaded diffusion schemes. We verify the effectiveness of OctFusion on the ShapeNet and Objaverse datasets and achieve state‐of‐the‐art performances on shape generation tasks. We demonstrate that OctFusion is extendable and flexible by generating high‐quality color fields for textured mesh generation and high‐quality 3D shapes conditioned on text prompts, sketches, or category labels. Our code and pre‐trained models are available at https://github.com/octree‐nn/octfusion .
Förderung
- National Key Research and Development Program of China
- Natural Science Foundation of Beijing Municipality
- National Natural Science Foundation of China
- Tencent
Zitieren
APA
Xiong, B., Wei, S., Zheng, X., Cao, Y., Lian, Z., & Wang, P. (2025). OctFusion: Octree‐based Diffusion Models for 3D Shape Generation. Computer Graphics Forum, 44(5). https://doi.org/10.1111/cgf.70198
Vancouver
Xiong B, Wei S, Zheng X, Cao Y, Lian Z, Wang P. OctFusion: Octree‐based Diffusion Models for 3D Shape Generation. Computer Graphics Forum. 2025;44(5). doi:10.1111/cgf.70198
BibTeX
@article{Xiong2025,
author = {Xiong, Bojun and Wei, Si‐Tong and Zheng, Xin‐Yang and Cao, Yan‐Pei and Lian, Zhouhui and Wang, Peng‐Shuai},
title = {OctFusion: Octree‐based Diffusion Models for 3D Shape Generation},
journal = {Computer Graphics Forum},
year = {2025},
volume = {44},
number = {5},
doi = {10.1111/cgf.70198}
}
EUVIMED