Point-StyleGAN: Multi-scale point cloud synthesis with style modulation

Yang Zhou, Cheng Xu, Zhiqiang Lin, Xinwei He, Hui Huang

Abstract: Highlights•We propose Point-StyleGAN, a generator adapted from StyleGAN2 architecture for multi-scale point cloud synthesis.•We introduce multi-resolution discriminators to overcome the under-constrained issue caused by the sparsity of points.•Experiments show that our method achieves competitive performance and much faster inference time than SOTA approaches.•An inversion encoder Point-pSp is also designed and applied to point cloud reconstruction, completion, and interpolation.