Vid-CamEdit: Video Camera Trajectory Editing with Generative Rendering from Estimated Geometry

TL;DR

Vid-CamEdit is a framework that enables realistic video re-synthesis along new camera paths by combining geometry estimation with generative rendering, even in challenging in-the-wild scenarios.

Contribution

It introduces a geometry-guided generative approach for camera trajectory editing that does not require extensive 4D training data.

Findings

Outperforms baselines in extreme trajectory extrapolation

Produces plausible videos from novel camera paths in real-world footage

Effectively handles in-the-wild videos with limited multi-view data

Abstract

We introduce Vid-CamEdit, a novel framework for video camera trajectory editing, enabling the re-synthesis of monocular videos along user-defined camera paths. This task is challenging due to its ill-posed nature and the limited multi-view video data for training. Traditional reconstruction methods struggle with extreme trajectory changes, and existing generative models for dynamic novel view synthesis cannot handle in-the-wild videos. Our approach consists of two steps: estimating temporally consistent geometry, and generative rendering guided by this geometry. By integrating geometric priors, the generative model focuses on synthesizing realistic details where the estimated geometry is uncertain. We eliminate the need for extensive 4D training data through a factorized fine-tuning framework that separately trains spatial and temporal components using multi-view image and video data. Our method outperforms baselines in producing plausible videos from novel camera trajectories, especially in extreme extrapolation scenarios on real-world footage.