Canonise ttD it doesn't work properly. It creates a mess,

[aneurinhunt]

Canonising a truncated truncated dodecahedron doesn't seem to work properly. Below is images showing Cn for each multiple of 100 from 0 to 10. This doesn't happen for other truncated archimedean solids (values in the thousands work fine). The last imagine is from http://dmccooey.com/polyhedra/CanonicalTruncatedTruncatedDodecahedron.html and is how I expect them to look after canonising them.

[huttarl]

Yeah that is quite a mess!
To be fair, the page does say "These operators ... are mainly intended for making the more traditional, convex polyhedra more symmetric". Whereas the initial ttD doesn't look convex.

Which is not to say the algorithm couldn't be improved. An algorithm that is stable across more, different types of shapes would be nice. I wonder if some kind of spring tension simulation (with edges as springs that pull) would help "relax" these concave polyhedra better. Although the polyhedron looks very close to convex at 300 iterations before it starts to go wonky, so maybe concavity is not the issue...