Representation of Triangle Meshes

Hi,

i'm currently searching for an adequate representation of triangle
meshes. I know that the choice of representation heavily depends on the
algorithms i want to run on the mesh. Currently I''m focusing on the
following two:

* meshes as indexed face lists (where each face (cell) stores pointers
or ids to vertices)
* meshes represented using half-edge or winged-edge data structure.

A lot of algorithms I deal with require fast access for neighborhood
queries or manipulating the topology of a mesh (edge collapse, edge
split,...). The half-edge data structure would be ideal as far as
neighborhood queries are concerned, but is it also ideal in the case of
removing faces or splitting/collapsing edges?

Some algorithms require access to (e.g) all edges in the mesh, so
should I rather explicitly store the edges in the mesh an let a
triangle consist of three edges instead of three vertices? Or should I
go for the half-edge data structure and convert all half-edges to
edges?

I'm somehow consfused on which way to go, so I'm asking you, the
community, for their experience with representations of triangle
meshes. Maybe you can shed some light on this topic..

thx in advance,
christoph

.

Relevant Pages

  • Re: n-ary representation and divisibility
    ... >Recently I have an observation about algorithms on divisibility of ... >But we perform all these algorithms mostly under decimal representation ... What about for different n-ary representation? ... The reason why a number in decimal form is divisible by 3 iff the sum ...
    (sci.math)
  • Re: hi everyone, i had s.ome questions about holographic information
    ... post-offset need to be stored for each base representation, ... i've not come across any algorithms like it, but as i can just see it ... To have a stream type compression with advanced algorithms (even ... since in this analogy the data set size is only limited by the memory ...
    (comp.compression)
  • Re: n-ary representation and divisibility
    ... > times the leftmost group and add on the second leftmost group, ... > But we perform all these algorithms mostly under decimal representation ... What about for different n-ary representation? ... > for m-ary representation of a. ...
    (sci.math)
  • Re: Little Endian -> Big Endian (Ada95 / GNAT), Whats with floating point types?
    ... The OP was converting out of and back into the same representation. ... The algorithms were both published in Proceedings of the ACM SIGPLAN ... How to Print Floating Point Numbers Accurately. ... You can find free implementations of these algorithms in C at ...
    (comp.lang.ada)
  • Re: Center of Contour
    ... in the boundary representation. ... triangle and the centroid of each triangle. ... weight is the SIGNED area of the triangle). ...
    (sci.image.processing)