Scalable visualization of massive point clouds

Graphical renderings of "raw" point clouds can be visually attractive. When combined, the many individual points and their attributes convey spatial structures, especially when viewed in a fluent 3D fly-through and on a stereoscopic display. For such interactive visualization, a sufficiently responsive graphics update is essential. Where many software packages support simple rendering of smaller (subsets of) point clouds, the size of point clouds currently acquired easily surpasses their rendering capabilities, even on a modern graphics card. We addressed this issue in recent experiments on the visual exploration of LiDAR (Light Detection And Ranging) datasets in our Virtual Reality systems. Out-of-core terrain rendering techniques are applied, which originate from graphic rendering engines used in games and flight-simulators. In this document some of these techniques and results are highlighted, and challenges in balancing rendering performance versus visual quality are discussed.

@Article { Haa09a,
  author       = "Haan, Gerwin de",
  title        = "Scalable visualization of massive point clouds",
  journal      = "Nederlandse Commissie voor Geodesie KNAW",
  volume       = "49",
  pages        = "59--67",
  year         = "2009",
  keywords     = "PBR, 3Di, GeoVis, AHN2",
  url          = "http://graphics.tudelft.nl/Publications-new/2009/Haa09a"
}