Globally Optimal Direction Fields
Abstract
We present a method for constructing smooth n-direction fields (line fields, cross fields, etc.) on surfaces that is an order of magnitude faster than state-of-the-art methods, while still producing fields of equal or better quality. Fields produced by the method are globally optimal in the sense that they minimize a simple, well-defined quadratic smoothness energy over all possible configurations of singularities (number, location, and index). The method is fully automatic and can optionally produce fields aligned with a given guidance field such as principal curvature directions. Computationally the smoothest field is found via a sparse eigenvalue problem involving a matrix similar to the cotan-Laplacian. When a guidance field is present, finding the optimal field amounts to solving a single linear system.
Additional Information
© 2013 ACM. This research was supported by a Google PhD Fellowship, the Hausdorff Research Institute for Mathematics, BMBF Research Project GEOMEC, SFB / Transregio 109 "Discretization in Geometry and Dynamics," and the TU München Institute for Advanced Study, funded by the German Excellence Initiative. Meshes provided by the Stanford Computer Graphics Laboratory and the AIM@SHAPE Shape Repository.Additional details
- Eprint ID
- 41016
- Resolver ID
- CaltechAUTHORS:20130829-154018080
- Google PhD Fellowship
- Hausdorff Research Institute for Mathematics
- BMBF Research Project GEOMEC
- SFB/Transregio 109 "Discretization in Geometry and Dynamics"
- TU München Institute for Advanced Study
- German Excellence Initiative
- Created
-
2013-08-29Created from EPrint's datestamp field
- Updated
-
2021-11-10Created from EPrint's last_modified field