CVMar 20, 2025

Repurposing 2D Diffusion Models with Gaussian Atlas for 3D Generation

arXiv:2503.15877v26 citationsh-index: 8
Originality Incremental advance
AI Analysis

This addresses the problem of 3D content creation for AI and graphics applications, offering a novel method but with incremental improvements in leveraging existing 2D models.

The paper tackles the challenge of 3D object generation by repurposing pre-trained 2D diffusion models, using a novel Gaussian Atlas representation and a large-scale dataset, achieving effective adaptation for 3D content generation.

Recent advances in text-to-image diffusion models have been driven by the increasing availability of paired 2D data. However, the development of 3D diffusion models has been hindered by the scarcity of high-quality 3D data, resulting in less competitive performance compared to their 2D counterparts. To address this challenge, we propose repurposing pre-trained 2D diffusion models for 3D object generation. We introduce Gaussian Atlas, a novel representation that utilizes dense 2D grids, enabling the fine-tuning of 2D diffusion models to generate 3D Gaussians. Our approach demonstrates successful transfer learning from a pre-trained 2D diffusion model to a 2D manifold flattened from 3D structures. To support model training, we compile GaussianVerse, a large-scale dataset comprising 205K high-quality 3D Gaussian fittings of various 3D objects. Our experimental results show that text-to-image diffusion models can be effectively adapted for 3D content generation, bridging the gap between 2D and 3D modeling.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes