PanoWan: Lifting Diffusion Video Generation Models to 360$^\circ$ with Latitude/Longitude-aware Mechanisms

Panoramic video generation enables immersive 360$^\circ$ content creation, valuable in applications that demand scene-consistent world exploration. However, existing panoramic video generation models struggle to leverage pre-trained generative priors from conventional text-to-video models for high-quality and diverse panoramic videos generation, due to limited dataset scale and the gap in spatial feature representations. In this paper, we introduce PanoWan to effectively lift pre-trained text-to-video models to the panoramic domain, equipped with minimal modules. PanoWan employs latitude-aware sampling to avoid latitudinal distortion, while its rotated semantic denoising and padded pixel-wise decoding ensure seamless transitions at longitude boundaries. To provide sufficient panoramic videos for learning these lifted representations, we contribute PanoVid, a high-quality panoramic video dataset with captions and diverse scenarios. Consequently, PanoWan achieves state-of-the-art performance in panoramic video generation and demonstrates robustness for zero-shot downstream tasks.

Authors

• Yifei Xia Peking University
• Shuchen Weng National Key Laboratory for Multimedia Information Processing, School of Computer Science, Peking University National Engineering Research Center of Visual Technology, School of Computer Science, Peking University
• Siqi Yang
• Jingqi Liu
• Chengxuan Zhu Peking University
• Minggui Teng Peking University
• Zijian Jia
• Han Jiang School of Software Engineering, Xi’an Jiaotong University
• Boxin Shi Peking University

Keywords

No keywords are indexed for this paper.