Ans: This is a research program, so I am sorry that it is not problem-free and it doesn't handle all the error well. Therefore, it will be great that if you screen capture the problem and send me an email with the data you are using, then I will check and let you know how to resolve it.
However, there is a common error in the inputs. The input models must be manifold, watertight, having same topology, and preferably free from degenerated and intersecting facets. Generally, my suggestion is to load the mesh into ReMESH (http://remesh.sourceforge.net/) to check for any geometric problems of your model. For example, I let ReMESH to fix the geometry problems at the time opening the model, remove the small shells, and close the holes, so that the model only has 1 shell and 0 boundary or genus. Although there are degenerated and intersecting facets (suggest to remove them as well in practice), cross-para can be run successfully.
2. How to place the feature points?
Ans: As mentioned in my papers, the feature points are used to construct the Voronoi diagram and build the base domain by tracing shortest paths. Therefore, they are preferred to be specified as uniform as possible, although this is not a necessary requirement. Particularly for the cylindrical shapes, which is quite ambiguous for path tracing, it is suggested that at least 2 or 3 feature points are specified along a circle to resolve the ambiguity. However, how to update the feature points to improve the mapping quality is case-by-case. It will be better that if you screen capture the problems or send me the files with a description telling where I should look at, then I can give you more specific advises. I am also working on a new method to place the feature points automatically, please stay tuned.
