Recammaster: Camera-Controlled Generative Rendering From a Single Video
Authors:
Top Cited Papers
- 19 October 2025
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 14834-14844
Abstract
Camera control has been actively studied in text or image conditioned video generation tasks. However, altering camera trajectories of a given video remains under-explored, despite its importance in the field of video creation. It is non-trivial due to the extra constraints of maintaining multiple-frame appearance and dynamic synchronization. To address this, we present ReCamMaster, a cameracontrolled generative video re-rendering framework that reproduces the dynamic scene of an input video at novel camera trajectories. The core innovation lies in harnessing the generative capabilities of pre-trained text-to-video models through a simple yet powerful video conditioning mecha-nism-its capability is often overlooked in current research. To overcome the scarcity of qualified training data, we construct a comprehensive multi-camera synchronized video dataset using Unreal Engine 5, which is carefully curated to follow real-world filming characteristics, covering diverse scenes and camera movements. It helps the model generalize to in-the-wild videos. Lastly, we further improve the robustness to diverse inputs through a meticulously designed training strategy. Extensive experiments show that our method substantially outperforms existing state-of-theart approaches. Our method also finds promising applications in video stabilization, super-resolution, and outpainting. Our code and dataset are publicly available at: https://github.com/KwaiVGI/ReCamMaster.Keywords
All Related Versions
Funding Information
- National Natural Science Foundation of China (62441615,62225603,U21B2004,62476241)
- Natural Science Foundation of Zhejiang Province (LZ23F020008)
This publication has 35 references indexed in Scilit:
- Global Structure-from-Motion Revisited2024
- Generative Camera Dolly: Extreme Monocular Dynamic Novel View Synthesis2024
- CoTracker: It Is Better to Track Together2024
- SparseCtrl: Adding Sparse Controls to Text-to-Video Diffusion Models2024
- Be-Your-Outpainter: Mastering Video Outpainting Through Input-Specific Adaptation2024
- Upscale-A-Video: Temporal-Consistent Diffusion Model for Real-World Video Super-Resolution2024
- Video stabilization: A comprehensive surveyNeurocomputing, 2022
- Kubric: A scalable dataset generator2022
- World-Consistent Video-to-Video Synthesis2020
- NTU RGB+D: A Large Scale Dataset for 3D Human Activity Analysis2016