文献类型：会议论文

英文题名：Multiple Level of Details Neural Implicit Surface Representation for Unconstrained Viewpoint Rendering

作者：Chen, Zhihua[1];Li, Yuhang[1];Dai, Lei[1];Li, Ping[2];Zhu, Lei[3];Sheng, Bin[4]

机构：[1]East China Univ Sci & Technol, Shanghai, Peoples R China;[2]Hong Kong Polytech Univ, Hong Kong, Peoples R China;[3]Hong Kong Univ Sci & Technol Guangzhou, Guangzhou, Peoples R China;[4]Shanghai Jiao Tong Univ, Shanghai, Peoples R China

会议论文集：IEEE Conference on Virtual Reality and 3D User Interfaces (VR)

会议日期：MAR 16-21, 2024

会议地点：Orlando, FL

语种：英文

外文关键词：Human-centered computing; Visualization; Visualization systems and tools; Visualization toolkits

摘要：To mitigate artifacts and performance degradation in neural implicit representation visualization for unconstrained viewpoint rendering, we propose a feature voxel grid-based neural representation architecture. This approach flexibly encodes the implicit surface with multiple levels of detail, facilitating high-quality rendering with dynamic switching between detail levels. Additionally, we implement a range-limited strategy to concentrate on sampling valid areas while excluding undefined areas. Our results in unconstrained viewpoint scenarios demonstrate the effectiveness of our method. This work extends the capabilities of neural implicit representations, broadening their potential applications beyond previously defined limitations.