This is an interesting process, but it notably didn't seem to produce any deserts, or really much in terms of long range texture variability. In a way it mimics AI language model output when you ask it to produce something original: superficially decent, but kind of meh and lacking in precise details when you start to really dig in.