Bending metal

JKayten, the first bend is interesting because the part you are bending appears to has a curve perpendicular to the bend you want to make. Often, I would flow along a curve setting a source and destination curve, but…

To make that look like it’s naturally bent I might rebuild the part in a bent shape by lofting to make surfaces, make a solid from the surfaces, and then wirecut and fillet to get the profile.

[To get better bends, I sometimes chop parts and rebuild the connecting areas entirely. Making Flowalongcurve Better for Sheetmetal "

Why do you feel you need to do all this? It seems like in general, you are far over-editing this stuff, and in the process you yourself are introducing errors that Rhino did not create in the first place.

As far as arc curves in general, please note that arcs over 90° must have more than 3 points. Between 90° and 180°, all arcs will have 5 points. This is totally normal and you cannot simplify it further. If you fillet something and it’s supposed to be an exact 90° corner and the arc has more than 3 points, the angle is actually more than 90°. Don’t fix the arc - fix the lines that you are filleting instead so that they are exactly 90°.


I’ve gone back and fixed the lines. As you can see from the images they all appear correct. I can then create a planar surface and extrude a solid. When go to make a hole, I get another mesh mess.

WTF2.3dm (117.5 KB)

Your outer curve still has a fundamental problem with it’s parameter values from -1.52508e+14 to -1.52508e+14 Parameter values should be have magnitudes of the size of the object or small, not essentially infinity.

From Properties Details:
Segment 2:
Open NURBS curve
start = (0,49.4728,112.153)
end = (0,6.47285,112.153)
degree = 2
control points: rational, count=31
knots: non-uniform, domain = -1.52508e+14 to -1.52508e+14
clamped at start and end
Curve is not smooth. If you explode it, it will become two or more curves.

By exploding the curve, and then splitting it at a corner I isolated the problem to the vertical straight line segment shown here:

This is probably one of the root causes of the problems you are having with this geometry. What are the exact steps and actions you used in creating the outline curve?

Good catch, I looked at it for awhile and didn’t see that. Reparameterize/Automatic also fixes the whole outer curve without having to explode everything. I would also extrude the curve with SplitAtTangents=Yes in this case.

Would definitely be good to know how the outer profile got that bizarre parametrization.


clement should get the credit for first finding the paremeterization issue in a post above.

Did you try _Reparameterize/Automatic on the edge curve? I did and the result still had the absurd values. I also tried Reparameterize specifying 0 and 100 and the start and end parameter values and also got the absurd values. Only when I used Explode and Split to isolate the straight section with the absurd values did Reparameterize work.

The straight section and the elliptical arc are parts of the same rational degree 2 curve which has a sharp corner. Explode does not separate them, and when are split the result is an elliptical arc and a straight section is a rational degree 2 curve. Rational means at least one of the control point weights is not 1.

The only ways I know to have a sharp corner in a degree 2 curve which will not Explode into 2 curves are:

a) Two control points at the same location. In that situation the number of reported control points for the curve is 1 greater than the number of visible control points. That is not the situation here.

b) Control point weight of 0. That is the apparent situation here but I don’t know how to check control point weight. A control point weight of 0 could result in a divide by zero which in turn could result in a value of around 10e+14.

So the cause of the strange parameterization is probably a control point weight of 0. How did that occur?

@JKayten How did you create the original version of the edge curve?

Added: Or if you created the corresponding surfaces first how did you create the original version of those surfaces?

I’ve been at this piece all day. It’s maddening.

The way I made the original profile was to draw lines with ortho on so they would be square and then applied fillets to the corners. When I need a line placed a certain distance from an edge, I’ll duplicate the edge and then use the gumball to enter the distance required. I then draw lines with the end snap to either end because I figure the guball has moved the line on a parallel axis to the original. I also create crossing lines by drawing them long past any snap point so the reman ortho and later trim them with each other.

LensPlate.3dm (890.9 KB)

Trying to finally merge all surfaces after exploding and rejooining will make an entire surface disappear in one instance. It’s difficult to give a detailed accouont of what’s taken place as I move along a path for a good distance until I run into some anomoly.

@JKayten You now have an invalid polysurface, which is probably due, at least in part, to the bad parameterization which was discussed above. Some other actions probably also contributed to it. Trying to build additional geometry using a fundamentally flawed foundation will only lead to more and more frustration.

Do you want to learn to create good models? If so you need to stop what you’re doing, back up and start over with the basics.

A good beginning would be to create the curve below without any problems and without having to make any corrections. Until you can do so there is no point in trying to press ahead hoping any problems will somehow go away.

Until you become competent in the basics of creating good curves you will thrashing and becoming more and more frustrated. After that are the basics of creating good simple surfaces, then the basics of blending surfaces and combining surfaces into more complex polysurfaces and solids.

I’m willing to help you learn the basics. Do you want to learn to model or do you want to thrash?

The rub is I think that’s what I’m doing. For instance as per the images below, the line you sdestermined as a problem is redrawn with ortho on from the end snap on the fillet edge. Is there any reason to believe the line drawn will not be verticle? Drawing the line past the point of the arc and then running intersect shows a point of intersection. so why then am I mistaken in my assumption?

Of course I don’t want to thrash. I’m really exhausted with these petty problems. I’m truly dismayed with the ongoing miniscule problems that seem impossible to actually generate.

By all means, I would appreciate your instruction. I’m anticipating a 2nd level course this may for the Mac version. Until then I’m taking instruction wherever I can find it.


I would add that I’m working on reproducing what is rather than creating somethin new. That’s by design because it creates demands that could otherwise be side stepped by a ‘design change’. I don’t mind a bit of thrashing as that often makes the learning sink deeper in. i have a nature that makes imprecision a genuine torment. And while I appreciate the well-meaning encouragement of those that suggest I go back to the basics, I reallly can’t stand spending time recreating pull toys and penquines.

I will however take your advice as well as any other’s because it’s a generosity and it’s well worth the attention.

Let’s back up. Where did the elliptical arc come from? How did you draw it?

The large elipitical arc was drawn with the arc tool using the end snaps to the horizontal and verticle curves. Afterwards it was deformed to match the item I’m trying to reproduce. I’m going to take a shot or two of the item and post them.

The order in which your are creating the curve is very different than how I would do it. I would create a basic shape, then add the tabs and holes, then fillet the corners.

Rhino is very precise. Lack of precision is usually due to user actions or decisions.

The fast way to creating good geometry is to spend the necessary time to learn the basics first.

Here’s some shots of what I’m working on.

I use a caliper to take measurements of the pieces and then draw them. I think in terms of straight lines and fillets. I use the arc tool where appropriate.

Looks like an interesting project.

When the camera was designed a set of “mechanical” drawings were created. Those drawings consisted mostly of straight lines, arcs and circles. Elliptical arcs would have been drawn with templates.

Exactly. I started as a draftsman when I was a kid using t-squares and triangles. I still approach design that way. That’s the prejudice I’m working with when I make assumptions on the lines I’m generating in Rhino.

The nice thing with the digital tools is the ability to modify infinitly without wearing out an eraser or vellum. No french curves required!

The arc used in this instance is not regular but altered to work with the view finder slots.

That’s what I can’t comprehend - how I’m somehow able to cause an error of that size - .014 degree - really?

I think he was referring to the out of tangency of the fillet section - which is clearly seen in the image in that post (#19). The lines leading into the fillet are probably exactly 90°, but the fillet corner control point does not line up with the tangent edges - which means that the fillet is not tangent. Somehow that middle control point got moved - if you used the fillet tool only, that shouldn’t happen.

BTW, what is your angle tolerance setting for the file? The angle tolerance setting determines what you consider sufficient to assume two curves or surfaces are tangent. By default it is 1°, which is absolutely huge IMO. My advice is to set it to 0.1° or less.


I use the application defaults. I’ve turned off ‘smart tracking’ and It’s making for clearer snappping. I’ll change the angle tolerance setting as recommended. Thanks