In this paper, point cloud data is used to construct a human body model, which in turn forms the transformation process of the human body. The corner point detection method is used to decompose the user-input two-dimensional garment shape contour lines, resulting in two types of data lines: shape lines and cross lines. Furthermore, the stereo configuration of the three-dimensional virtual garment is automatically generated using the void volume modeling method and displayed in real-time. The real graphic effect of 3D style is also studied by applying lighting and material processing methods. And take the actual case as the starting point to complete research on the use of 3D clothing design. After analysis, it can be seen that the standard deviation of point cloud data measuring human body parts is small, and the average errors of thickness, circumference, and volume are 1.21%, 2.64%, and 0.053%, respectively. The experimental study designed four garment models with the highest chest fit in terms of T-shirt styles’ fit measurement.