---
base_model:
- alibaba-pai/Wan2.1-Fun-V1.1-1.3B-Control-Camera
license: cc-by-nc-sa-4.0
pipeline_tag: image-to-video
---

<a href="https://arxiv.org/abs/2511.18886v1"><img src='https://img.shields.io/badge/arXiv-2511.18886-red?style=flat&logo=arXiv&logoColor=red' alt='arxiv'></a>&nbsp;
<a href="https://vivocameraresearch.github.io/magicworld/"><img src='https://img.shields.io/badge/Project-Page-Green' alt='GitHub'></a>&nbsp;
<a href="https://creativecommons.org/licenses/by-nc-sa/4.0/deed.en"><img src='https://img.shields.io/badge/License-CC BY--NC--SA--4.0-lightgreen?style=flat&logo=Lisence' alt='License'></a>&nbsp;
<a href='https://youtu.be/OB_eVa_qIIg'><img src='https://img.shields.io/youtube/views/OB_eVa_qIIg'></a>

# MagicWorld: Interactive Geometry-driven Video World Exploration

MagicWorld is an interactive video world model that generates dynamic scene evolution from a single image based on user actions. It autoregressively synthesizes continuous scenes by integrating 3D geometric priors (Action-Guided 3D Geometry Module - AG3D) for structural consistency and a History Cache Retrieval (HCR) mechanism to mitigate error accumulation over multi-step interactions.