Offset Curve Not Coming Out

[Cartierusm]

OK, so I have a surface that is the carved top of my guitar. I then do this Curve > Curves from Objects > Duplicate Border. This gives me an outline curve of the carved top, matching it's ups and downs on the edge. Now when I go Curve > Offset Curve > .19, it gives me an exact outline at the correct distance from the original BUT the outline is broken up into about 4 pieces and where the pieces meet they overlap with points that don't line up, very jagged.

So how do I make an exact copy of the curve but at an even distance away from the original so I can make a binding channel. Thanks.

[RandK]

Offset should work on a good curve. Try to fitcrv tolerance 0 degree 3 the dup border curve. That should create a new curve to very tight tolerance that you can then offset. Fitcrv, rebuild etc will make small changes to the curve so check to make sure it still matches your purpose.

[chmillman]

Offset is not necessarily reliable in 3D, especially if you have kinks or other probelms with your original. If you could post just the curve as an attachment here, someone may be able tolook at it and help you. --ch

[Cartierusm]

I can't have any changes it has to be perfect.

Here is the curve by itself.

[chmillman]

Your original curve is pretty junky. In the attached file, I offset it, it is almost OK, but there is one problem spot (where the points are). This is due to a kink in your original curve. Offset is CPlane dependent, so offset from the top view.

I adjusted the curve slightly to remove the kinks and make it offset better, you can try that one if you want. The deviation is around .001. I would think about improving your original curve and surface geometry if possible... --ch

[Cartierusm]

Thanks, I'll check it out. Why is the original junky? I just cleaned up the original in Illustrator, then I Imported it into Rhino, then make an extrusion and made a carved top, then cut the carved top to the outline of the body, then made the current curve as I described abovel. Is that method producing inferior results?

[chmillman]

Well, the Illustrator curves may have been poor, depending on how you made them. Illustrator doesn't necessarily guarantee you smooth curves, there's no good tangency or curvature control at joints between sections.

Good curve geometry is the key to making good surface geometry in Rhino (or any other program, for that matter). If you really want to make smooth geometry, you need to be careful how you make stuff, with the fewest control points possible and looking at the curvature graph while editing.

The curve you are trying to offset is also an edge curve, that is, it's derived from the intersection of two surfaces. That means it's always going to be somewhat more noisy than the original curves, and if your originals aren't that clean, the derived stuff will be much worse.

You might try creating your curves in Rhino as a comparison. I think you should be able to get a close approximation of your shape with around 30 control points and have it be reasonably smooth. --ch

[Cartierusm]

CH, thanks, but how did you do that, I need to be able to create them myself becasue the offset changes periodically. Thanks.

[RandK]

Your curve has 1044 points (pointson will show them) and if you explode it 19 segments. Try a analyze>curve>curvature graph (this is 3d). The spikey stuff are curve control points that are not tangent to the surrounding ones or at an abrupt change in the direction of the curve. As ch says you need to simplify the original curves and the offset which will also usually be overly complex. fitcrv is one tool. At 0 tolerance it uses your absolute tolerance units setting so it's probably much smaller than the eye can possibly see.

Binding curves/routes etc are hair-tearing learning experiences. Don't forget to think about that piece of plastic binding and the X-Y radius you can bend it to (heat will help) around the body. Any itsy-bitsy radius change in your curve will turn into a gap because you can't force the plastic to conform to it. Binding doesn't bend well in the Z direction at all but you can force it to a smooth gradual ramp up/down. When you offset the edge by the cutter_radius - binding thickness, the top surface at the inside of the binding ledge may be doing something different than that body edge. Also think about the contact points of the endmill that are going to run along cutting that ledge. The CAM has to build a toolpath that gets close but doesn't gouge so it's going to leave a little extra material. Bottom line is the design for manufacturability may need a little tweaking. What the eye can see trumps aerospace precision Make test parts and adjust.

[Cartierusm]

Weird in Illustrator I only have about 10 points. For the future, although I appreciate the help, stuff like pointson doesn't mean anything to me...LOL I'm pretty proficient getting around Rhino but terms and tags are beyond me right now...please dumb it down.

So Rand, what you're saying is if I click on the curve (before making the offset) > then typing in fitcrv at 0 will help clean up the curve?

[RandK]

Dup border is giving you a curve that reflects the complexity of the surfaces. As ch said an intersection of freeform surfaces is going to be complicated

PointsOn and curvature graph are useful tools for inspecting and by dropping/dragging/deleting/adding points changing curves.

Try selecting your original curve. PointsOn; Curvaturegraph. See lots of points and the curvature "hairs" are trying to show you what the curvature is doing. Now do your fitcrv and see how it reduces the number of points and really cleans up the curvature graph. You won't visually see a difference in the simplified curve. Now you can do your offsets. Hint: you probably want to do an overcut on the binding ledge leaving a 1/16th or so and then a quick cut to the full offset to prevent tearout. So do two offsets one slightly oversized for the overcut.

[Cartierusm]

Thanks, I'll try that later tonight.