[extremelearning]

One reason/situation where the Fibonacci method is preferred is because it is a direct construction method, which can be coded in a few lines, rather than an indirect iterative method. The second is that because an energy minmization method is minimizing the sum of forces, it more closely minimizes average distance between points, rather than absolute minimum distance which is what packing distance focuses on.

As I describe in the article, different methods produce similar but slightly different solutions. An optimal solution for one objective function, may not be the optimal for a different objective function. I then give details about how the solution that optimizes volume of the convex hull is different to the solution that optimizes for packing distance, etc.

[phkahler]

One advantage is that you can use arbitrary N.

If you just want N in a certain range, we can use the triangle-based polyhedron and successively quadruple or triple the number of faces. Then use the face normals as points. This gives visually appealing distributions without any real oddities.