libnurbs - a C++ Non-Uniform Rational B-Splines Library

Non-Uniform Rational B-Splines (NURBS) curves and surface are very general mathematical surfaces widely used for representing complex three dimensional shapes in computer graphics.

The goal of libnurbs is to provide a clean, robust and powerful library with the ability to define, manipulate, and analyze NURBS curves and surfaces. We will be building off of the foundation provided by the openNURBS library, implementing functionality missing from that library and making changes as needed. The goal of the openNURBS effort is increased interoperabilty between various CAD systems, so they have no incentive to develop or release a more full-featured library (that's Rhino, their commercial platform) - hence the need for this project.

This project is distinct from the old NURBS++ project previously developed on this page - both the people and the code are different. The previous web content is available here and previous NURBS++ releases are still available from the files section of the project website. The license of the original NURBS++ code has not changed - it remains LGPL. The SVN repository contains the NURBS++ codebase (including some fixes for more modern compilers). The intent of the new libnurbs effort (at least for now) is to use the Modified BSD license, and thus the NURBS++ code won't be directly usable. (If the decision is made to switch to LGPL then merging in code becomes possible.) New development is being done in the Git repository.

Broad Project Goals

The overall plan is fairly vague at this stage, but the immediate steps are clear (development takes place on the project website):

• Convert the header documentation in openNURBS to Doxygen format. This is a large and unexciting job, but it's also something virtually anyone can help with and necessary for the long term - openNURBS is a large code base and the header comments constitute the primary documentation.
• Work towards building openNURBS with strict compiler flags. A CMake build system is already in place.
• Identify what features are not present (doxygen conversion will help with that) and determine a "what do we need first" dependency layout. It is highly probable that surface trees and curve trees will be first on the list.
• Start implementing features. There are a number of relevant research papers pertaining to things like surface triangulate and NURBS intersection algorithms that will be useful as guides - a list of links to those papers is also a TODO item.

Related projects

There are actually a number of other open source projects pertaining to NURBS, some of them quiet impressive. Most are incompatibly licensed for the purposes of libnurbs, but they deserve mention as possible resources for "end users" not constrained by licensing:

• SINTEF SISL Library is the most mature and feature complete open source NURBS library currently available, although it does not seem to have gained a following (as yet) in the open source community. It is GPL licensed. Recent versions build using CMake. SINTEF has other interesting NURBS related code available on their site, also GPL licensed.
• Ayam is an open source NURBS modeler based on Tcl/Tk and OpenGL. It is BSD licensed (some features require GPL libraries) and it is the most promising candidate for features/code that can be re-used (or re-implemented, anyway) in libnurbs.