Simplification

From Meshlab

Simplification filters are filters that get in input a mesh and build another mesh composed by a smaller number of triangles. There are three simplification filters inside MeshLab

• Clustering. Based on a simple vertex clustering approach. A uniform grid is used to collapse all the vertices falling in the same grid cell onto a single vertex. It is a very fast algorithm but not very accurate, a bound of the introuduced error is half of the diagonal of each grid cell. Topological inconsistencies are quite probable that will be created (non manifold edges). This is a typical drawback of any clustering-based simplification algorithms. In our implementation the only parameter that can be specified by the user is the length of the side of the cubic grid cell. This length can be introduced either as an absolute value in the space of the mesh itself, or as a percentage of the diagonal of the bounding box of the object. With this kind of approach you cannot know in advance the final size of the mesh, but only knowing what could be at the end the final average edge length (e.g. approx. the same of the grid cell side).

• Quadric edge collapse. A variant of the well known edge collapse algorithm based on quadric error metric proposed by Michael Garland and Paul Heckbert . The user can control the quality of the simplification through some parameters:
  • Quality threshold for penalizing edge collapses that creates bad shaped faces.
    The value is in the range [0..1] (0 a degenerate triangle 1 a equilateral one); the value of this parameter must be interpreted as follows:
    • 0 accept any kind of face (no penalties),
    • k penalize faces with quality than k, proportionally to their shape quality.
  • Normal cheking for penalizing edge collapses that flip the normal of the original surfaces (it can help to avoid some nasty artifacts that can arise in very flat areas).
• Quadric edge collapse with texture coordinates preservation