UltraShape 1.0: High-Fidelity 3D Shape Generation via Scalable Geometric Refinement

TL;DR

UltraShape 1.0 introduces a scalable 3D diffusion framework that generates high-fidelity 3D geometries through a two-stage process, enhanced data processing, and a novel localized refinement technique, achieving high-quality results with limited training data.

Contribution

The paper presents a new scalable 3D diffusion method with a two-stage pipeline, innovative data processing, and a localized geometric refinement approach, improving 3D shape quality from publicly available datasets.

Findings

Achieves high geometric quality comparable to state-of-the-art methods.

Develops a comprehensive data filtering and hole-filling pipeline.

Demonstrates effective localized refinement using voxel queries and RoPE encoding.

Abstract

In this report, we introduce UltraShape 1.0, a scalable 3D diffusion framework for high-fidelity 3D geometry generation. The proposed approach adopts a two-stage generation pipeline: a coarse global structure is first synthesized and then refined to produce detailed, high-quality geometry. To support reliable 3D generation, we develop a comprehensive data processing pipeline that includes a novel watertight processing method and high-quality data filtering. This pipeline improves the geometric quality of publicly available 3D datasets by removing low-quality samples, filling holes, and thickening thin structures, while preserving fine-grained geometric details. To enable fine-grained geometry refinement, we decouple spatial localization from geometric detail synthesis in the diffusion process. We achieve this by performing voxel-based refinement at fixed spatial locations, where voxel queries derived from coarse geometry provide explicit positional anchors encoded via RoPE, allowing the diffusion model to focus on synthesizing local geometric details within a reduced, structured solution space. Our model is trained exclusively on publicly available 3D datasets, achieving strong geometric quality despite limited training resources. Extensive evaluations demonstrate that UltraShape 1.0 performs competitively with existing open-source methods in both data processing quality and geometry generation. All code and trained models will be released to support future research.