Researchers have introduced a novel 3D human generation method called Chupa, which combines the generative capabilities of diffusion models with neural rendering techniques to create diverse and realistic 3D human figures. The core innovation of Chupa lies in generating high-quality human meshes from SMPL-X grids. It consists of two sub-tasks: 2D normal map generation and 3D reconstruction based on the normal map, which can be applied to various human poses, demonstrating excellent quality.