| （英） |
We focus on mean, variance, and quaternion of 3D Gaussian Splatting (3DGS) and aim to generate CGH faster than the point light source method without the need for multi-view images. The mean, variance, and quaternion of each 3DGS correspond to its position, size, and rotation, which we call geometric quantities. We present a method that can compute each CGH pattern pixel by pixel in order to support parallelization for higher speed and memory saving operation, |