cs.CVNov 4, 2026

RDT-3D: Resolution-Decoupled Training for High-Fidelity 3D Geometry Generation

Authors: Jiale Zhang, Yang Liu, Chengyang Li, Jin Yuan, Qianxi Jia, Taicheng Huang

Organizations: JD.com, Beijing, China

Abstract

High-fidelity 3D geometry generation requires maintaining global structural coherence alongside fine-grained surface details. While recent coarse-to-fine frameworks leverage Rotary Position Embeddings (RoPE) to condition local queries, we reveal a critical training-inference resolution discrepancy: coordinate guidance and geometric supervision are aligned during training, whereas inference decodes high-resolution geometry from coarse spatial coordinates. This mismatch introduces a severe distribution shift that degrades fine details. To overcome this, we propose Resolution-Decoupled Training (RDT), which anchors RoPE guidance at 512 resolution while supervising geometry at 1536, enabling robust detail extrapolation. Additionally, to eliminate high-resolution data bottlenecks, we present Dual3D-300K, a large-scale paired dataset comprising 300K watertight 3D point clouds explicitly aligned across 512 and 1536 resolutions. Extensive experiments demonstrate that RDT substantially improves sharp geometric regions, while Dual3D-300K offers a scalable benchmark for high-resolution 3D generation.

Figures & tables

Explore similar work

CardsList
  1. VoxLution: Evolving Voxels for High-Fidelity 3D Scene Generation

    Oct 30, 2026Daoyi Gao, Xinyi Zhang, Alexey Bokhovkin +23D Scene GenerationIndoor Scene Generation