Information

  • Publication Type: Invited Talk
  • Workgroup(s)/Project(s):
  • Date: 2009
  • Event: University of Erlangen Research Seminar
  • Location: Erlangen, Germany
  • Conference date: 27. July 2009 –

Abstract

Real-time rendering is a quickly developing field in computer graphics. Recent advances in graphics hardware make it possible to tackle completely new challenges, and to rethink old ones. While previously, the main focus of real-time rendering lay on classical problems like visibility and level-of-detail rendering, nowadays we see new challenges in the form of interactive procedural content generation, handling of massive amounts of data, and interactive simulation of extremely complex objects.

In this talk, I will cover some of the recent advances we had in our group. First, we try to integrate procedural modeling techniques with the new parallel programming paradigms made commonly available through modern GPUs, and map L-system generation onto hardware to accelerate the generation of large L-systems. Then, I'll briefly show some results for really large scale visualization and editing of a huge point-based model consisting of over 1.2 Billion point samples of a Roman catacomb. Finally, I will treat a new approach to handle the classical visibility problem, where we show how to calculate visibility of a whole scene by exploiting the spatial coherence of visibility, thus accelerating the process so it becomes viable for interactive scene design.

Additional Files and Images

No additional files or images.

Weblinks

No further information available.

BibTeX

@talk{WIMMER-2009-ARTR2,
  title =      "Advances in Real-Time Rendering",
  author =     "Michael Wimmer",
  year =       "2009",
  abstract =   "Real-time rendering is a quickly developing field in
               computer graphics. Recent advances in graphics hardware make
               it possible to tackle completely new challenges, and to
               rethink old ones. While previously, the main focus of
               real-time rendering lay on classical problems like
               visibility and level-of-detail rendering, nowadays we see
               new challenges in the form of interactive procedural content
               generation, handling of massive amounts of data, and
               interactive simulation of extremely complex objects.  In
               this talk, I will cover some of the recent advances we had
               in our group. First, we try to integrate procedural modeling
               techniques with the new parallel programming paradigms made
               commonly available through modern GPUs, and map L-system
               generation onto hardware to accelerate the generation of
               large L-systems. Then, I'll briefly show some results for
               really large scale visualization and editing of a huge
               point-based model consisting of over 1.2 Billion point
               samples of a Roman catacomb. Finally, I will treat a new
               approach to handle the classical visibility problem, where
               we show how to calculate visibility of a whole scene by
               exploiting the spatial coherence of visibility, thus
               accelerating the process so it becomes viable for
               interactive scene design. ",
  event =      "University of Erlangen Research Seminar",
  location =   "Erlangen, Germany",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2009/WIMMER-2009-ARTR2/",
}