The -r switch reconstructs a previously generated mesh. The mesh is read from a .node and an .ele file, and possibly a .face file. If a .face file exists, it is read and used to constrain subfaces in the mesh, else, TetGen will automatically identify the subfaces, internal subfaces are also identified by comparing the attributes of two adjacent tetrahedra. Subsegments will be automatically identified from the existing subfaces. The reconstructed mesh is distinguished from its origin with a different iteration number.

For example, ``tetgen -r xxx.1'' reads the mesh in files xxx.1.node, xxx.1.ele and possibly xxx.1.face and xxx.1.edge if they exist; reconstructs the mesh; outputs it into three files xxx.2.node, xxx.2.ele and xxx.2.face. Now, you can give xxx.2 as input in the above command, you will get another mesh saved files xxx.3.node, and so on. Mesh iteration numbers allows you to create a sequence of successively finer meshes. They also allow you to produce a sequence of meshes using error estimate-driven mesh refinement.

You can refine a mesh by combining -r with -q, -a, and -i switches. Several ways are possible:

• You can impose tighter quality constraints by using -q with a smaller number, or -a followed by a smaller volume than you used to generate the mesh you are refining.
• You can create a .vol file, which specifies a maximum volume for each tetrahedra, and use the -a switch (without a number following). Each tetrahedron's volume constraint is applied to that tetrahedra.
• You can create a .node file, which contains a list of additional nodes that you want them to be in the mesh. Use -i switch to tell TetGen that you have an additional node list needs to be inserted.

-r is not compatible with -p and -I, don't try to use them together.