Vid2World: Crafting Video Diffusion Models to Interactive World Models

TL;DR

Vid2World introduces a novel method to transform pre-trained video diffusion models into interactive world models, enhancing their controllability and applicability in complex decision-making environments.

Contribution

The paper proposes a systematic approach to adapt video diffusion models into interactive world models with causalization and action guidance mechanisms.

Findings

Effective transfer of video diffusion models to interactive environments

Improved action controllability in world models

Successful application across robotics, gaming, and navigation domains

Abstract

World models, which predict future transitions from past observation and action sequences, have shown great promise for improving data efficiency in sequential decision-making. However, existing world models often require extensive domain-specific training and still produce low-fidelity, coarse predictions, limiting their usefulness in complex environments. In contrast, video diffusion models trained on large-scale internet data have demonstrated impressive capabilities in generating high-quality videos that capture diverse real-world dynamics. In this work, we present Vid2World, a general approach for leveraging and transferring pre-trained video diffusion models into interactive world models. To bridge the gap, Vid2World systematically explores video diffusion causalization, reshaping both the architecture and training objective of pre-trained models to enable autoregressive generation. Additionally, it incorporates a causal action guidance mechanism to enhance action controllability in the resulting interactive world models. Extensive experiments across multiple domains, including robot manipulation, 3D game simulation, and open-world navigation, demonstrate that our method offers a scalable and effective pathway for repurposing highly capable video diffusion models into interactive world models.