What question did this study set out to answer?

This research aims to explore the recent advancements in 3D virtual human reconstruction and animation using Neural Radiance Fields and Gaussian Splatting methods.

May 20, 2026Journal of Computer-Aided Design & Computer Graphics0 citationsOpen Access

A Survey on 3D Virtual Human Reconstruction and Animation Based on Neural Radiance Fields and Gaussian Splatting

JWJunchao Wang JZJinshan Zhang LYLujia Yang

Key Points

This research aims to explore the recent advancements in 3D virtual human reconstruction and animation using Neural Radiance Fields and Gaussian Splatting methods.
Systematic review of techniques for 3D virtual human reconstruction and animation.
Focus on data-driven methods and dynamic modeling through NeRF and 3DGS.
Analysis of challenges in cross-modal temporal biases and dynamic modeling efficiency.
NeRF improves geometric consistency in single-view reconstructions of complex poses.
3DGS enhances efficiency in non-rigid deformation modeling with real-time rendering capabilities.
Identified challenges in temporal alignment across modalities and trade-offs between modeling efficiency and fidelity.

Abstract

随着元宇宙、数字孪生与智能交互方法的快速发展，3D虚拟人作为连接物理世界与数字空间的媒介，其高保真重建与自然驱动方法成为计算机视觉与人工智能交叉领域的核心挑战。文中聚焦神经辐射场（NeRF）与3D高斯溅射（3DGS）方法，系统地梳理了3D虚拟人重建与驱动方法的前沿进展，在形式上遵循从数据驱动重建到内容生成的常见流程，同时以动态建模、多模态对齐和物理约束融合三大维度为线索展开分析。在重建方法层面，重点探讨NeRF与3DGS的范式创新：基于NeRF的方法通过隐式连续函数建模动态辐射场，在单视角重建中实现复杂姿态的几何一致性优化；3DGS方法则凭借显式可编辑性与实时渲染能力，显著地提升非刚性形变建模效率。在驱动方法层面，围绕音频与视频两大模态，分类阐述唇部同步、全脸表情生成、多模态情感驱动等关键方向，解析了时序建模网络、物理引导的神经渲染架构的优化路径。最后归纳该方法当前面临的跨模态时序偏差、动态建模效率-保真度矛盾等挑战，并展望物理引导学习、轻量化架构设计等未来发展方向，旨在为虚实融合的数字生态构建提供方法参考与理论支撑。

Ask AI

Helpful

Bookmark

View Full Paper

Cite This Study

Wang et al. (2026) studied this question.

synapsesocial.com/papers/6a0d4efcf03e14405aa9a29b https://doi.org/10.3724/sp.j.1089.2025-00267

Ask AI

Helpful

Bookmark

View Full Paper

May 20, 2026Journal of Computer-Aided Design & Computer Graphics0 citationsOpen Access

A Survey on 3D Virtual Human Reconstruction and Animation Based on Neural Radiance Fields and Gaussian Splatting

JWJunchao Wang JZJinshan Zhang LYLujia Yang

Key Points

This research aims to explore the recent advancements in 3D virtual human reconstruction and animation using Neural Radiance Fields and Gaussian Splatting methods.
Systematic review of techniques for 3D virtual human reconstruction and animation.
Focus on data-driven methods and dynamic modeling through NeRF and 3DGS.
Analysis of challenges in cross-modal temporal biases and dynamic modeling efficiency.
NeRF improves geometric consistency in single-view reconstructions of complex poses.
3DGS enhances efficiency in non-rigid deformation modeling with real-time rendering capabilities.
Identified challenges in temporal alignment across modalities and trade-offs between modeling efficiency and fidelity.

Abstract

Ask AI

Helpful

Bookmark

View Full Paper

Cite This Study

Wang et al. (2026) studied this question.

synapsesocial.com/papers/6a0d4efcf03e14405aa9a29b https://doi.org/10.3724/sp.j.1089.2025-00267

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

Ask AI

Helpful

Bookmark

View Full Paper