HD-VGGT: High-Resolution Visual Geometry Transformer
Tianrun Chen, Yuanqi Hu, Yidong Han, Hanjie Xu, Deyi Ji, Qi Zhu, Chunan Yu, Xin Zhang, Cheng Chen, Chaotao Ding, Ying Zang, Xuanfu Li, Jin Ma, Lanyun Zhu

TL;DR
HD-VGGT is a dual-branch transformer architecture that enables efficient, high-resolution 3D reconstruction by combining coarse global predictions with detail refinement, while suppressing unstable features.
Contribution
It introduces a novel dual-branch design and feature modulation technique to improve high-resolution 3D reconstruction efficiency and robustness.
Findings
Achieves state-of-the-art reconstruction quality on high-resolution images.
Effectively suppresses unreliable features in ambiguous regions.
Reduces computational costs compared to full-resolution transformer models.
Abstract
High-resolution imagery is essential for accurate 3D reconstruction, as many geometric details only emerge at fine spatial scales. Recent feed-forward approaches, such as the Visual Geometry Grounded Transformer (VGGT), have demonstrated the ability to infer scene geometry from large collections of images in a single forward pass. However, scaling these models to high-resolution inputs remains challenging: the number of tokens in transformer architectures grows rapidly with both image resolution and the number of views, leading to prohibitive computational and memory costs. Moreover, we observe that visually ambiguous regions, such as repetitive patterns, weak textures, or specular surfaces, often produce unstable feature tokens that degrade geometric inference, especially at higher resolutions. We introduce HD-VGGT, a dual-branch architecture for efficient and robust high-resolution 3D…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
