Collaborative Video Diffusion: Consistent Multi-video Generation with Camera Control

TL;DR

This paper introduces Collaborative Video Diffusion, a method that generates multiple consistent videos from different camera trajectories, enabling more controllable and coherent multi-view video synthesis.

Contribution

The paper proposes a novel cross-video synchronization module with epipolar attention, improving multi-video consistency in camera-controlled video generation.

Findings

Significantly improved multi-video consistency over baselines

Effective synchronization of frames from different camera trajectories

Demonstrated scalability with state-of-the-art camera control modules

Abstract

Research on video generation has recently made tremendous progress, enabling high-quality videos to be generated from text prompts or images. Adding control to the video generation process is an important goal moving forward and recent approaches that condition video generation models on camera trajectories make strides towards it. Yet, it remains challenging to generate a video of the same scene from multiple different camera trajectories. Solutions to this multi-video generation problem could enable large-scale 3D scene generation with editable camera trajectories, among other applications. We introduce collaborative video diffusion (CVD) as an important step towards this vision. The CVD framework includes a novel cross-video synchronization module that promotes consistency between corresponding frames of the same video rendered from different camera poses using an epipolar attention mechanism. Trained on top of a state-of-the-art camera-control module for video generation, CVD generates multiple videos rendered from different camera trajectories with significantly better consistency than baselines, as shown in extensive experiments. Project page: https://collaborativevideodiffusion.github.io/.