Free Viewpoint Rendering
Free-viewpoint rendering (FVR) aims to generate realistic images of a scene from arbitrary viewpoints, going beyond the limitations of the original captured perspectives. Current research heavily utilizes neural radiance fields (NeRFs) and related architectures, often incorporating techniques like multi-resolution hash encoding and spatio-temporal sampling to improve efficiency and handle dynamic scenes, including moving humans and objects. These advancements are driving progress in applications such as virtual and augmented reality, immersive telepresence, and high-quality digital avatar creation, particularly by enabling photorealistic rendering from sparse or monocular video inputs.
Papers
HQ3DAvatar: High Quality Controllable 3D Head Avatar
Kartik Teotia, Mallikarjun B R, Xingang Pan, Hyeongwoo Kim, Pablo Garrido, Mohamed Elgharib, Christian Theobalt
FlexNeRF: Photorealistic Free-viewpoint Rendering of Moving Humans from Sparse Views
Vinoj Jayasundara, Amit Agrawal, Nicolas Heron, Abhinav Shrivastava, Larry S. Davis