Progressive Rendering of Massive Point Clouds in WebGL 2.0 Compute

Information

• Publication Type: Bachelor Thesis
• Workgroup(s)/Project(s):
  • Rendering and Modeling
• Date: October 2019
• Date (Start): July 2019
• Date (End): October 2019
• Matrikelnummer: 01526299
• First Supervisor: Markus Schütz
• Keywords: point cloud, progressive rendering, webgl

Abstract

Rendering large point clouds is a computationally expensive task, and various optimizations are required to achieve the desired performance for realtime applications. It is typical to store the point data hierarchically to enable fast retrieval and visibility testing in point clouds that consist of billions of points. However, rendering the selected nodes is still a demanding task for the graphics units on modern devices. Especially on mobile devices rendering millions of points every frame is often not possible with sufficient frame rates. Techniques that progressively render the points of a point cloud were proposed to reduce the load on the GPU. The results of the previous frames are recycled, and details are accumulated over multiple frames. Combining hierarchical structures with progressive rendering, therefore, houses an exciting opportunity for increasing the performance for massive point clouds.

This work investigates a novel approach to render massive point clouds progressively in the browser by transforming the hierarchical structure locally into an unstructured pool of points. The pool is then rendered progressively with compute shaders and continuously updated with new nodes from the octree.

Additional Files and Images

Weblinks

• https://github.com/ifl0w/potree

BibTeX

@bachelorsthesis{Rumpler-2019-PPC,
  title =      "Progressive Rendering of Massive Point Clouds in WebGL 2.0
               Compute",
  author =     "Wolfgang Rumpler",
  year =       "2019",
  abstract =   "Rendering large point clouds is a computationally expensive
               task, and various optimizations are required to achieve the
               desired performance for realtime applications. It is typical
               to store the point data hierarchically to enable fast
               retrieval and visibility testing in point clouds that
               consist of billions of points. However, rendering the
               selected nodes is still a demanding task for the graphics
               units on modern devices. Especially on mobile devices
               rendering millions of points every frame is often not
               possible with sufficient frame rates. Techniques that
               progressively render the points of a point cloud were
               proposed to reduce the load on the GPU. The results of the
               previous frames are recycled, and details are accumulated
               over multiple frames. Combining hierarchical structures with
               progressive rendering, therefore, houses an exciting
               opportunity for increasing the performance for massive point
               clouds.   This work investigates a novel approach to render
               massive point clouds progressively in the browser by
               transforming the hierarchical structure locally into an
               unstructured pool of points. The pool is then rendered
               progressively with compute shaders and continuously updated
               with new nodes from the octree.",
  month =      oct,
  address =    "Favoritenstrasse 9-11/E193-02, A-1040 Vienna, Austria",
  school =     "Research Unit of Computer Graphics, Institute of Visual
               Computing and Human-Centered Technology, Faculty of
               Informatics, TU Wien ",
  keywords =   "point cloud, progressive rendering, webgl",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2019/Rumpler-2019-PPC/",
}