Virtual Reality Experience Analysis from Point Cloud Data

This study explores the implementation of point cloud visualization in virtual reality environments to evaluate immersion based on point cloud characteristics. Two visualization methods were utilized, including points and meshes, and the method was tested on point clouds from three popular datasets: Paris-Carla-3D, Toronto-3D, and Stanford 3D Indoor Scene. The point-based visualization better preserved the original point cloud’s geometries and colors, but visualization through surfaces formed by points added greater realism to the experience. However, the illumination options of the points resulted in a loss of realism. Mesh visualization modified both the geometry and colors, producing models that were less realistic than those created by the point visualization. Occlusions affected the realism of all scenes, causing a loss of information and influencing the generation of erroneous illumination. Immersion was hindered by the large number of points displayed on the screen simultaneously, which significantly reduced Frames Per Second. Despite this, no problems related to Virtual Reality sickness were observed beyond a brief initial adaptation period.