We now no longer randomly split, but always split on the axis that is larger.
This seems to produce much, much better mazes; random choice, even with extra parameters, resulted in ugly mazes with very long passages that did not resemble other recursive division mazes.
We now no longer randomly split, but always split on the axis that is larger.
This seems to produce much, much better mazes; random choice, even with extra parameters, resulted in ugly mazes with very long passages that did not resemble other recursive division mazes.