Information

  • Publication Type: Bachelor Thesis
  • Workgroup(s)/Project(s):
  • Date: October 2014
  • Date (Start): 21. October 2013
  • Date (End): 28. February 2014
  • Matrikelnummer: 1026408
  • First Supervisor: Stefan OhrhallingerORCID iD
  • Keywords: sampling, mesh, modeling

Abstract

Currently, in modeling there is a lot of manual work involved in reducing the number of vertices/triangles in geometry to balance efficiency vs. accuracy of the intended shape. Our aim is to automatically find an optimum for that, by resampling an input model, e.g. created by an artist.

We need to process the geometry of a model discretely (as in points/triangles instead of a continuous boundary) for rendering it on the screen, animation or collision detection. Such models can be given by an explicit boundary (spline, triangle mesh) or implicitly by a signed distance function, e.g. defined at octree nodes, from 3D scanner data approximated by implicit surface reconstruction methods, such as Poisson.

We want to represent this boundary by as few points as possible while losing only as much detail as determined by a user-controlled parameter. We use a medial condition which adapts to local features of arbitrary size. This also permits us to drop connectivity (edges, triangles), so that the output becomes a point cloud. The parameter can control either the maximum deviation of the generated boundary to the input, or a point budget.

Additional Files and Images

Additional images and videos

Additional files

thesis: Bachelor Thesis thesis: Bachelor Thesis

Weblinks

No further information available.

BibTeX

@bachelorsthesis{geyer_lukas-2014-bat,
  title =      "Volume-aware Mesh Resampling",
  author =     "Lukas Geyer",
  year =       "2014",
  abstract =   "Currently, in modeling there is a lot of manual work
               involved in reducing the number of vertices/triangles in
               geometry to balance efficiency vs. accuracy of the intended
               shape. Our aim is to automatically find an optimum for that,
               by resampling an input model, e.g. created by an artist.  We
               need to process the geometry of a model discretely (as in
               points/triangles instead of a continuous boundary) for
               rendering it on the screen, animation or collision
               detection. Such models can be given by an explicit boundary
               (spline, triangle mesh) or implicitly by a signed distance
               function, e.g. defined at octree nodes, from 3D scanner data
               approximated by implicit surface reconstruction methods,
               such as Poisson.  We want to represent this boundary by as
               few points as possible while losing only as much detail as
               determined by a user-controlled parameter. We use a medial
               condition which adapts to local features of arbitrary size.
               This also permits us to drop connectivity (edges,
               triangles), so that the output becomes a point cloud. The
               parameter can control either the maximum deviation of the
               generated boundary to the input, or a point budget. ",
  month =      oct,
  address =    "Favoritenstrasse 9-11/E193-02, A-1040 Vienna, Austria",
  school =     "Institute of Computer Graphics and Algorithms, Vienna
               University of Technology ",
  keywords =   "sampling, mesh, modeling",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2014/geyer_lukas-2014-bat/",
}