Nanyang Technological University collaborated with the SenseTime S-Lab team to develop an efficient 3D human body modeling framework called GauHuman, which is based on Gaussian Splatting technology and draws on the concept of Human NeRF. GauHuman has made significant breakthroughs in both speed and rendering quality. It only takes 1-2 minutes to complete the modeling and can perform real-time rendering at 189 frames per second, far exceeding the current technical level.
The S-Lab team of Nanyang Technological University and SenseTime recently released an exciting research result, launching an efficient 3D human body modeling framework based on Gaussian Splatting, named GauHuman. This framework has achieved significant breakthroughs in rapid reconstruction and real-time rendering, providing an efficient solution for human body modeling in the digital field. GauHuman completes modeling in 1 to 2 minutes and renders up to 189 frames per second, surpassing existing solutions and having broad application prospects, covering games, movies, virtual reality and other fields. Through SMPL parameters and neural networks, GauHuman effectively processes monocular human body videos and quickly generates high-quality 3D digital humans, creating a new situation in 3D human body modeling. GauHuman's modeling framework is based on Gaussian Splatting and inspired by the previous Human Neural Radiation Field (Human NeRF).
The emergence of GauHuman marks a new level of 3D human modeling technology. Its efficiency and high precision will greatly promote the digitalization process of games, movies, virtual reality and other industries, providing creatives with more powerful tools and More convenient workflow. In the future, as the technology continues to mature, GauHuman is expected to be widely used in more fields.