抄録
Interactive mesh editing techniques typically encode geometric shapes using discrete differential equations and approximately solve them using the least squares method. However, these techniques fail to precisely preserve user-defined constraints. In our method, user-defined constraints, such as positional constraints, are treated as hard constraints, and then the combination of hard and soft constraints is converted into a typical least-squares system using QR decomposition. Our method stably and efficiently calculates vertex positions that preserve differential properties. A well-known problem with hard constraints is over-constraints. If over-constraints are involved, they may cause the solver to halt the computation. We show how our framework resolves over-constrained situations, which include redundant and conflicting constraints.
