The best way to solve this problem with an increasing number of splines

Question

I have a different number of splines representing text.
My goal is to ensure that the number of splines is always not lower than the threshold I need.

If I'm given 3 splines, but I only need 3 splines, then everything is great and I don't need to do anything.

But if I'm given 5 splines and I need 6 splines, then I just convert this to mesh, and select a random edge. And then I split the edge, and turn the result back into splines.
After which I get the required 6 number of splines.

If instead of 6 splines I need 10 splines, then I make 5 cuts instead of 1.

Since the number of cuts can be large, it is important that they are in random places for beauty.

But... there is one problem.

If a random cut falls on a "symbol with a hole" (an edge between two different splines, not between the same one), or if all these splines are symbols with holes,

Then, after a random cut, instead of increasing the number of splines, their number will decrease.

In order for their number to really increase, an 2 additional cut will have to be made.

In total we come to the conclusion "number of cuts $\ne$ goal splines $-$ current splines".

So I had to just use a repeat zone, something like

Continue randomly cutting until get lucky and the quantity is right.

Well... Is there a way to solve this without a repeat zone? Any math tricks or other tricks? Something that will work faster than repeat zone on a large number of splines?

Answer 1

...Oh, I love this task, so I played around with it a bit...

First of all:
You got it absolutely right: The Fill Curve node is extremely helpful here, as it reliably prepares the geometry in such a way that there are no more internal curves.

This seems to have solved the actual problem.

---

Nevertheless, I don't want to withhold my gimmick from you, as it might work a little more efficiently than your previous approach:

Essentially, I also use Fill Curve as a starting point to process the problematic areas correctly.

However, I then divide the mesh into two parts:

1. the outer (original) curves
2. the inner edges, which I obtain by triangulating the mesh

The advantage here, however, is that I edit the existing curves directly and thus avoid having to convert them into a mesh again.

To obtain the inner edges, I use the following group:

_{Extract interior edges}

I use these inner edges to cut the curves, which looks like this within the Repeat Zone:

_{Process the cuts}

But the trick here is to separate the selected edge from the rest so that Random Value does not select the same edge again in the next run.

This process also takes place in a group that looks like this:

_{Select a random edge from the interior edges}

By then checking which points of the outer curves intersect with the selected edge, I can separate the curve that I want to cut.

The Accumulate Field node then helps me to separate the points that lie on one side or the other of the intersection.

_{Cut the curve into two parts}

By defining with Set Spline Cyclic that these curves should be closed, I finally get the necessary curves, which I then only have to join with Join Geometry.

Depending on the target value, this process is then repeated until the desired number of curves is reached.

_{(Blender 4.1.0+)}

Answer 2

Brilliantly! N-gons mode from Fill Curve helped to unambiguously determine the number of partition iterations.

Since polygons in Blender cannot contain holes, converting to n-gons automatically solves problems with letters that contains holes.

This is the same trick I was talking about. How come I didn't think of this right away?.

Now, instead of entering a large number of iterations (ex. 1111 cycles),
you can find them out using "number of cuts $=$ goal $-$ face count from n-gones"

But... this method requires two Fill Curves nodes, which does not improve performance much. So still puzzled.