Ok I try to explain it without any math:
#0 (CONSTANT)
Rings which have endorsements of unchanging size grater than its carrying body nearly keep their price no matter how big their ringdiameter get.
Your Moonball rings are of this class, this explains why you expierienced almost no price variations.
I initially thought they would be of class #2 (see there).
#1 (LINEAR)
Rings which doesn't change thickness and width doubles their price when they double their diameter.
#2 (QUADRATIC)
Rings with doesn't change thickness but do change width
quadruples their price when they double in diameter.
[The volume of a spherical hull with constant thickness does it too, it quadruples when ball diameter doubles. So if your Moonballs would grow with ringsize (this is what I initially thought) but keep their wall thickness, your rings would be of this class too.]
#3 (CUBIC)
Rings which do change thickness and their width octuplicates (!!) their price when they double in size.
Thats the case for my Haskell ring, in the form I have it now.
And the resaon why Shapeway's maximum volume/price requirement is a problem for me.
As it is now the logo is exactly as wide as the outer ring diameter.
It's important for me that the ring should look the same (same proportions) no matter of the finger diameter. I could do different scalings for logosize but that woud screw proportions and bring me at best from #3 to #2.
The symbols run into the ring main body tangentially in the center.
(This turned out accidentaly but I'am especially happy with that)
Keeping this poses additional constrain to scaling.
Sidenote: Different scaling of different parts in my models does reqire changing only a few numbers in my code. It's very straightforward in most cases.
I would use the cocreator-size-ranges solely for previewing an average / most-probable price for women & men but I doesn't want to prevent a women from buying a wide or a man from buying a slim ring, so I guess I have to split each of woman & man size ranges into two ranges making effectively four cocreator models for one kind of ring.
[Sadly the (haskell) programming community consists almost entirely out of men.]
But you do have a point, making the hight parameterizable as a second cocreator option is a must
because it's already a single number is in my code.