Briefly let's discuss performance. One unusual aspect of this algorithm
is that extreme simplification is faster. That is reducing a model to
fewer elements can be completed in less time than a lesser reduction.
Why is that? Well the initial value for E, the allowable boundary error
is larger for more extreme simplifications. More edge collapses will be
allowed in the initial iterations of the algorithm and the slower
optimization to reach the final tetrahedral count is performed over
fewer elements.
|