| title | GeoFix: Geometry Fixing for 4D Consistent Generation |
|---|---|
| authors | Yeobin Hong, Hyojun Go, Dohun Lee, Jeongjae Lee, Sunghwan Hong, Federico Tombari, Zhaochong An, Jong Chul Ye, Konrad Schindler, Dominik Narnhofer |
| date | 2026-10-09 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2610.14293 |
| upvotes | 23 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf |
| abstract | Video diffusion models increasingly serve as priors for 4D generation, but the generated videos often lack geometric consistency. Recent approaches address this through post-training by designing geometric rewards, yet this can degrade texture and fine detail, requiring auxiliary appearance rewards and careful reward balancing. To address this challenge, we propose GeoFix, a unified framework that improves geometric consistency and visual quality with a single geometric reward, applicable to both training and inference-time reward alignment. Our reward measures cross-view depth consistency in static scene regions, targeting structural inconsistencies without directly constraining appearance or penalizing valid object motion. Crucially, depth agreement increases with camera field-of-view overlap even in real videos and 4D datasets, allowing a smaller camera movement to be mistaken for better geometry. We account for this effect by calibrating the reward against the overlap-dependent trend. GeoFix achieves a 39% improvement in WorldScore 3D consistency and a 4% increase in VBench Quality Score. |
cite
@misc{hong2026geofix,
title={GeoFix: Geometry Fixing for 4D Consistent Generation},
author={Yeobin Hong and Hyojun Go and Dohun Lee and Jeongjae Lee and Sunghwan Hong and Federico Tombari and Zhaochong An and Jong Chul Ye and Konrad Schindler and Dominik Narnhofer},
year={2026},
eprint={2610.14293},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2610.14293}
}