Papers
arxiv:2607.09125

4D Human-Scene Reconstruction from Low-Overlap Captures

Published on Jul 10
· Submitted by
Sangmin
on Jul 14
Authors:
,
,
,
,

Abstract

Existing volumetric capture of dynamic human performance achieves high fidelity with dense camera arrays. However, in real-world scenarios, only a handful of low-overlap cameras are available, which degrades the output quality and leaves large areas unobserved. Recent 4D reconstruction methods have focused on low-overlap settings, yet they still produce noticeable artifacts in under-observed regions. Video diffusion models have emerged as another option, but they show geometrically inconsistent results for humans. To address these limitations, we propose StudioRecon, a pipeline that reconstructs 4D human scenes from sparse, low-overlap cameras by decoupling background and humans. We densify background supervision by synthesizing hundreds of camera-controlled novel views with a video diffusion model. We also robustly initialize deformable Gaussian humans with cross-view identity association and triangulated multi-view keypoint fitting. Finally, our recursive enhancement module with motion-adaptive consistency injection harmonizes the composed output, thereby further avoiding remaining artifacts. We achieve state-of-the-art novel view synthesis across four real-world datasets and demonstrate applications such as novel trajectory rendering and human replacement.

Community

Four cameras around a room, roughly 90° apart, with barely any overlap between neighboring views. COLMAP doesn't even register them. That's the setting we went after.

Our insight: backgrounds and humans want different priors, so we stop making one model solve both.

🎬 Video diffusion densifies the background, turning 4 real views into hundreds.
🧍 SMPL constrains the humans, where video diffusion falls apart under motion.
✨ A recursive enhancement module harmonizes the two, without per-frame flicker.

Across 8 scenes from EgoHumans, Harmony4D, Mobile Stage, and SelfCap, StudioRecon outperforms prior methods on every scene: +1.5 to +5.0 dB PSNR over the best baseline, with LPIPS reduced by 33 to 74%.

Accepted to SIGGRAPH Conference Papers '26. First two authors contributed equally.

Amazing work! Thanks for sharing

Sign up or log in to comment

Models citing this paper 0

No model linking this paper

Cite arxiv.org/abs/2607.09125 in a model README.md to link it from this page.

Datasets citing this paper 0

No dataset linking this paper

Cite arxiv.org/abs/2607.09125 in a dataset README.md to link it from this page.

Spaces citing this paper 0

No Space linking this paper

Cite arxiv.org/abs/2607.09125 in a Space README.md to link it from this page.

Collections including this paper 1