ReCapture: Generative Video Camera Controls for User-Provided Videos using Masked Video Fine-Tuning
This work addresses the need for controllable camera motion in user-provided videos, offering a novel solution for video editing and generation applications.
The paper tackles the problem of generating new videos with novel camera trajectories from a single user-provided video, which existing methods cannot handle, and achieves this by using a two-step approach involving multiview diffusion models and masked video fine-tuning to produce clean and temporally consistent results.
Recently, breakthroughs in video modeling have allowed for controllable camera trajectories in generated videos. However, these methods cannot be directly applied to user-provided videos that are not generated by a video model. In this paper, we present ReCapture, a method for generating new videos with novel camera trajectories from a single user-provided video. Our method allows us to re-generate the reference video, with all its existing scene motion, from vastly different angles and with cinematic camera motion. Notably, using our method we can also plausibly hallucinate parts of the scene that were not observable in the reference video. Our method works by (1) generating a noisy anchor video with a new camera trajectory using multiview diffusion models or depth-based point cloud rendering and then (2) regenerating the anchor video into a clean and temporally consistent reangled video using our proposed masked video fine-tuning technique.