Paper Overview
Field: Computer Vision (CV) Authors: Zizun Li, Haoyu Guo, Runzhe Teng Published: 2026-05-26 arXiv: 2505.21448
Summary
Camera-controlled video generation has achieved remarkable progress in recent years. However, existing video-to-video re-rendering methods primarily rely on Supervised Fine-Tuning (SFT) using synthetic datasets. At present, synchronized, multi-view real-world video data is extremely scarce. Consequently, the prevailing paradigm often exhibits limited generalization when processing out-of-distribution real-world videos, with models struggling to accurately adhere to physical scales and camera trajectories.
To bridge this gap, the authors propose Geo-Align, the first Reinforcement Learning framework specifically designed for camera-controlled video re-rendering. Built upon a pretrained model, the model is optimized through a scale-aware perceptual reward mechanism.
Key Contributions
- Metric geometry reward: A metric 3D estimator extracts precise camera trajectories from generated videos, explicitly penalizing rotation and translation deviations.
- Paired-data-free pipeline: A data pipeline strategy built on real-world conditioned videos and target camera trajectories derived from synthetic data removes the dependency on paired data.
Results
Extensive experiments show that Geo-Align consistently outperforms existing supervised learning baselines in precise camera controllability and visual fidelity, demonstrating the effectiveness of the approach.
---
*Auto-collected on 2026-05-26.*