At a high level, the framework is divided into three parts - a flexible a halfedge mesh data structure, an optimized linear algebra package (based on Eigen), and code for various geometry processing algorithms. Each algorithm comes with its own viewer for rendering.
Detailed documentation and unit tests for each of these parts can be found in the docs and tests directories of this repository.
We're just getting rolling here, so stay tuned for more! :-)
Since geometry-processing-js already implements many of the fundamental operations needed for geometry processing, it's easy to get up and running very quickly. Here's a short snippet showing how to solve a Poisson equation on a mesh loaded by the GUI, which uses built-in routines for constructing the Laplace and mass matrices:
// assign an index to each vertex of the mesh let vertexIndex = indexElements(geometry.mesh.vertices); // build cotan-Laplace and mass matrices let A = geometry.laplaceMatrix(vertexIndex); let M = geometry.massMatrix(vertexIndex); let rhs = M.timesDense(rho); // solve Poisson equation with a given right-hand side rhs let llt = A.chol(); let phi = llt.solvePositiveDefinite(rhs);
Clone the repository and change into the projects directory
git clone https://github.com/geometrycollective/geometry-processing-js.git cd geometry-processing-js/projects
Open the index.html file in any of the sub directories in a browser of your choice (Chrome and Firefox usually provide better rendering performance than Safari).
Dependencies (all included)
Linear Algebra - A wrapper around the C++ library Eigen compiled to asm.js with emscripten. Future updates will compile the more optimized sparse matrix library Suitesparse to asm.js. (Note that this wrapper can also be used for other, non-geometric projects which seek to use Eigen on the web; you can find the standalone release here)
Rendering - three.js