S-DyRF: Reference-Based Stylized Radiance Fields for Dynamic Scenes

• DOI: 10.48550/arxiv.2403.06205
• Publication Date: 2024-03-10
• Authors: Xingyi Li; Zhiguo Cao; Yizheng Wu; Kewei Wang; Ke Xian; Zhe Wang; Guosheng Lin
• Keywords: Stylized fact

Current 3D stylization methods often assume static scenes, which violates the dynamic nature of our real world. To address this limitation, we present S-DyRF, a reference-based spatio-temporal method for neural radiance fields. However, stylizing scenes is inherently challenging due to limited availability stylized reference images along temporal axis. Our key insight lies in introducing additional cues besides provided reference. end, generate pseudo-references from given These facilitate propagation style information entire scene. For coarse transfer, enforce novel views and times mimic details at feature level. preserve high-frequency details, create collection pseudo-rays pseudo-references. serve as detailed explicit guidance achieving fine transfer. Experiments on both synthetic real-world datasets demonstrate that yields plausible results space-time view synthesis scenes.