Photo
Modeling with Subdivision Surfaces
Overview Our work on modeling with subdivision surfaces is part of a joint project between IBM, NYU, and Dassault Systemes. The primary goal of this project is to explore the applicability of subdivision techniques to interactive surface editing for shape design.

cut-and-paste editing Cut-and-paste editing Our most recent work is related to the design and implementation of an interactive cut-and-paste tool for surfaces. The basic idea is similar to image cutting and pasting, in which a source region of an image is selected, extracted, and pasted at a target location in the same or in another image. In the case of surfaces, complications arise from the need to define what part of a surface constitutes detail to be transferred and what part is the base with respect to which the details are expressed. In addition to addressing this issue, we developed algorithms for target area identification, for establishing a map between the source and target regions involved in pasting, and we paid careful attention to efficiency, to allow for interactive manipulation of the pasted feature on the target surface. Read more.

sharp features Sharp features & trimming We developed a method for creating sharp features and trim regions on multiresolution subdivision surfaces along a set of user-defined curves. The novelty of our work lies in the ability to create sharp features and trim boundaries anywhere on a surface and in the fact that the resulting representation remains within the multiresolution subdivision framework. Preserving the original representation has the advantage that other operations applicable to multiresolution subdivision surfaces can subsequently be applied to the edited model. Read more.

variational design Variational design We have also investigated the issue of interactively deforming an initial geometric shape to obtain a new one that satisfies certain requirements. These requirements are formulated as a set of constraints and the underlying geometric representation is modified to meet these constraints. We have designed and implemented a prototype variational modeler that allows interactive editing of complex objects of arbitrary topology. We take advantage of our underlying multiresolution representation to allow editing at different scales. Most importantly, we preserve original detail during editing in a natural way.