Real-Time Continuous Level of Detail Rendering of Point Clouds

Real-time rendering of large point clouds requires acceleration structures that reduce the number of points drawn on screen. State-of-the art algorithms group and render points in hierarchically organized chunks with varying extent and density, which results in sudden changes of density from one level of detail to another, as well as noticeable popping artifacts when additional chunks are blended in or out. These popping artifacts are especially noticeable at lower levels of detail, and consequently in virtual reality, where high performance requirements impose a reduction in detail. We propose a continuous level-of-detail method that exhibits gradual rather than sudden changes in density. Our method continuously recreates a down-sampled vertex buffer from the full point cloud, based on camera orientation, position, and distance to the camera, in a point-wise rather than chunk-wise fashion and at speeds up to 17 million points per millisecond. As a result, additional details are blended in or out in a less noticeable and significantly less irritating manner as compared to the state of the art. The improved acceptance of our method was successfully evaluated in a user study.

1 Introduction

Point-cloud rendering poses a variety of challenges, such as the enormous amount of points that is necessary to represent a model in similar detail as textured meshes, as well as aliasing artifacts due to poor scan quality and the lack of mipmapping. One of the most common sources of point-based models are 3-dimensional scans of real objects, buildings and even whole countries. Smaller objects and structures may consist of a few million coloured points in millimeter resolution, whereas scans of large areas and whole countries may consist of hundreds of billions of points. Figure 1:

Left: Discrete LOD structure with sudden jumps in density and popping artifacts under motion. Middle: Screenshot of our method in virtual reality. No visible seams across different levels of detail due to a continuous reduction in density, and also continuously reduced density towards the periphery to reduce geometric complexity where less is needed. Endeavor point cloud courtesy of NVIDIA [4]. Right: Continuous transition from one LOD to another.