Advice on how to fill and oddly shaped area

I need to create [a] surface[s] to fill in this oddly shaped area.

The red curves indicate the end of the surface where other surfaces meet at right angles. The purple curves indicate the end of the surface where it mates up continuously with the adjacent surface. The blue curves are tangent with the adjacent surface but are flexible in their position.

This illustrates that surface. Here I am dealing with the inside (dark area of the starboard propeller) of the twin keel:

Obviously, this is an oddly shaped piece of steel bent into shape. If I use the PATCH command I can visualize a surface that is roughly what I want. The problem is that (at least within my skill) the surface produced does not match the outlines within the tolerance need to join the surfaces at the edges.

If I use NETWORKSRF, EDGESRF, or SWEEP2 I get something like this with excess bends:

My question: How can I fill this area is something nearly as smooth as what Patch shows but matching the edges within the tolerance?

Problem Filling.3dm (3.0 MB)

and why dont you create 3 surfaces then? having edgy curves to create a smooth geometry will not work out. i thought we had this issue already but maybe i dont fully get in which orientation you might want to have it, i can only guess for now because on that image i dont see such harsh continuity breaks. and maybe post those curves that one can help you further.

Three surfaces is what i expect would have to happen. The problem is to get them without the kinks that show up at the bottom.

obviously you have created curves which do not admit any smooth geometry translation. from a screen shot which shows your surface from above i cant say much more. if those curves are highly confidential then ok maybe try another screen shot :smiley: with points turned on for example.

Sorry, I made the files to attach but forgot to add that one. Updated.

From what I see, it should be BB-61 USS Iowa.
There are some problems with these curves, they inevitably affect the surfaces.
I do not know the method by which curves have been obtained (photos, laser scanners, photogrammetry). If you can make changes, you can bring the result home … otherwise, if you can not modify, the surfaces will show some discontinuities.
A good starting point might be to rebuild the curves (see enclosed file, crv color cyan))

Problem Filling.3dm (2.6 MB)

The curves came from tabular data.

As already written with those curves you can not make a surface or pole surface clean. There are curves that have continuity G 0. Moreover, these surfaces I imagine connect to other surfaces … so it is difficult to figure out how to intervene. From the picture I can only try to guess which surface it is.

The surface in question is the inside of what you have circle. move a little to the right into the dark spot and you have it.

As you say, there is no way that these surfaces can be clean. My objective is simply to get them as clean as possible. I’d settle for something that approaches what patch does. Someone in 1939 bent metal plates into odd shapes.

I am going to cheat a little on some of these cures but I am stuck with the general shape.

is that square plate on top of another surface? or is it the actual surface boundary? if its just added onto another surface you can build the surface below far smoother and trimm into it whatever you need straight. its difficult to understand what you really need here if you dont give us some tables or a part of the original blue prints. if you have numbers only then its going to be even trickier of course. if you can cheat a little and correct the surrounding surfaces according this one then here you go, maybe this is ok for you.

Problem Filling-2.3dm (2.7 MB)

The red curves are at the edge of the plate. There are surfaces that join at right angles. The purple curves are where there rest of the hull joins.

The attached file has the adjoining surfaces. Plus, I’ve cheated a bit on the lower curve and cleaned it up.
Problem Filling.3dm (3.4 MB)

the problem is that you are trying to define a surface through crosstangential curves. which are not only geometrically difficult but are also produced very bad. for instance the red curves are chopped into ugly pieces which do not allow any smooth surface let alone that you have a duplicate curve. rebuild or refitting any of this will cause a change. while you have taken data from some tables they do not describe what the surface actually wants to be and where it wants to go. blindly trying to fit any curves together ist probably not the best choice. maybe we can find a strategy to get you authentic surfaces by understanding the actual geometry of it. its at least the best i can do now.

for example how would that part look like when you make a closed surface, can you draw a sketch?
by defining this very edge and understanding what will happen we might be able to figure what we can do to.

a good chance would be to understand where the surface would want to head that we just trimm it off exactly at its boundaries. an untrimmed smooth surface will not be possible without changing (understanding) it.

I had been distracted by real work. I got this pretty well broken down like this.

The problem is that the two lower triangles look horrible (but the top one looks good enough).

Any idea who to fill in the triangular areas without bulging like that?

The problem here is that this is the end of the ship and things have to mate up along the red lines. They are fixed in place.

Problem Plating II.3dm (6.1 MB)

There are big problems with curves and surfaces. The approach I propose is in line with the model’s quality. Maybe he can help.

Problem Plating II.3dm (8.5 MB)

What are you doing here?


And yes, they are weird curves. I wonder how they came up with something like this in the days before computers.

Reissbretter :relaxed:


Please understand that I do not write and speak a good English. What I mean is that you are trying to get surfaces from bad curves and, for that reason, the result will never be good. Also note that when IOWA was designed “in pencil” and not computer (and this is the case for anything) the sheets may have been cut following the drawings, but then the assembly did not create a perfect curvature. For this reason, I wrote that they are bad curves, meaning they are bad Rhino curves. (Multiple knots, G0 continuity, curves with too many checkpoints, they can only generate problems.) I’m not here to criticize and apologize if this message came in. I just say that you should improve the curves so that your work on Rhino Gives you more satisfaction.


I understand. This is a case where I am trying to find the least worst solution to the problem. Because it is a prominent area and because the folk who will use this a rivet counters, I am stuck with the shape. Clearly it is never going to create a perfect shape.

I have cheated a little since the original post by smoothing out some of the curves where I don’t think people will notice.

In the “old days” they would have lofted the surface with untrimmed edges, extending the surface edges past the trimmed boundary curves. They then created the curve using the intersection of two or more surfaces. It’s still a pretty common practice today.

Exactly. Start with large smooth surfaces to define the form. Then trim out the panels required. This method is just trying to take the pee out of the pool water.