120 likes | 212 Views
Explore the algorithm and steps involved in mesh fixup, including vertex and edge manipulation using specific tolerances. Learn about vertex and edge merging, edge breaking, and drawbacks like spike generation. Discover the importance of user intervention and careful tolerance selection to maintain features. Understand the challenges of backtracking and radiosity requirements for continuous interpolation. Dive into processing stages such as Grouper, Lsect, Pmesh, and Emesh, with emphasis on spatial subdivision and solidity values. Uncover common problems like missing polygons and future work in object recognition and procedural correction.
E N D
Mesh fixup II Rudro Samanta Princeton University
Using tolerances • Give vertices and edges “tolerance” • Steps in the algorithm • Vertex merging • Edge merging • Edge breaking
Drawbacks • Spike generation • User intervention required • Tolerance must be chosen carefully • How to maintain features ? • Backtracking is expensive
Radiosity requirements • Continuous interpolation • Well shaped • Sufficient radiator density • Sufficient receiver node density • Intersecting faces should be represented explicitly
Processing stages • Grouper • Merges geometry • Removes T vertices • Maximizes coplanar faces
Stages continued • Lsect • Removes face intersections by breaking planes • Coplanar • Removes overlapping coplanar faces • Pmesh • Uniformly subdivides faces • Initially squares and then traiangulation • Emesh
Problems • Again a error tolerance is required • Authors say this is automatically computed (“scene relative tolerance”)
Spatial subdivision • Divide space by polygon planes • Generate solidity values • Create consistent model
Drawbacks • Missing polygons • May generate spurious features • Connected parts • Solidity determination becomes unclear
Conclusion • Approaches are usually straightforward • Need some user intervention in many pf the processes (usually for error bound) • Missing polygons still a mystery
Future work • The next level • (Fuzzy ?) Object recognition • Characteristic database and detection • Procedural correction