Human NeRF
Human NeRFs leverage neural radiance fields to create realistic 3D models of humans from images or videos, aiming for accurate novel view and pose synthesis. Current research focuses on improving efficiency through techniques like hypernetworks and multi-input/multi-output architectures, as well as enhancing generalizability by incorporating visibility awareness, pre-trained models, and handling challenging scenarios such as interacting hands and mirror reflections. This technology has significant implications for applications like virtual and augmented reality, motion capture, and 3D animation, offering more efficient and realistic human representation than traditional methods.
Papers
HyperPlanes: Hypernetwork Approach to Rapid NeRF Adaptation
Paweł Batorski, Dawid Malarz, Marcin Przewięźlikowski, Marcin Mazur, Sławomir Tadeja, Przemysław Spurek
Efficient Dynamic-NeRF Based Volumetric Video Coding with Rate Distortion Optimization
Zhiyu Zhang, Guo Lu, Huanxiong Liang, Anni Tang, Qiang Hu, Li Song