3d Curve Full Radius?

[tlehnhaeuser]

Can anyone figure out how to put a full radius on the endpoints of the 3D Curve

 

Created in V9

fullrad.ics

[Guest IronKevin]

The easiest way would be to create a 2D Shape of it the create the 3D Curve from that.

 

IK

[tlehnhaeuser]

Kev,

that won;t work because the 2 line segments are not planar.

tom

[Guest IronKevin]

I that case I don't see an easy solution

 

IK

[Guest mwalls]

Tom,

 

Have you tried putting a 3d shape, such as a cylinder, at each end of the 2 lines you want to join, making sure they are perpendicular. Then use the 3d curve tool to put in the radius between the shapes like you would do to run a wire or something. Then you could extract that radius for your use. I haven't tried it yet, but am just thinking out loud.

 

Mike

[tlehnhaeuser]

Tom,

 

Have you tried putting a 3d shape, such as a cylinder, at each end of the 2 lines you want to join, making sure they are perpendicular. Then use the 3d curve tool to put in the radius between the shapes like you would do to run a wire or something. Then you could extract that radius for your use. I haven't tried it yet, but am just thinking out loud.

 

Mike

14891[/snapback]

 

Mike,

yes but the problem I have is finding the exact midpoint and centerpoint of the radius between the two endpoints.

See what happens is actually it is a full radius that "twists" in the middle to maintain tangency to each line segment.

 

What I have done and may be acceptable ( not sure yet) is create a line segment from each point and filleted each end. It is not a full radius but may get buy with it. I had to extend the line segments first and then measure the distence between the 2 endpoints and divide in half.

 

Which brings up n ER: the ability to use 3D smartDimensions on 3D curves and points. It doesn;t work and I had to creat a block tand resize to get the measurement between the two points.

 

Thanks

Tom

 

[Guest mwalls]

Tom,

 

One more thought, can you start out by making the three curves all planer first, and then manipulate the ends to create what you want?

 

Mike

[tlehnhaeuser]

Nice thought.

Heres what happens. I can't get the full radius unless I use the connector option and then it creates "extension lines" which I can't maintain tangency.

 

Then I tried your approach using a 2D shape but when I create a 3D curve and go to move a point with the full rad on it I lose tangency to lines.

 

It was a good thought though.

Tom

 

[Mike Allen]

I'm not clear on what "Full Radius" means. Does it mean that the radius is a critical dimension & can't be changed to meet the lines? Are the lines supposed to be tangent to the curved section, or are they meant to connect at an angle as you have shown? You definitely have one line that doesn't project tangent to the given arc. As I see it, the given radius won't connect tangent to both lines (unless you change the length or position of at least one of the lines), even with a twist in the middle, but it appears to be possible for another radius. Which is critical - the radius of the curve, or the length/position of the lines?

 

Given the ability to change the length of one of the lines & also make the curve tangent to both lines, I came up with the following solution. I took a block (for a projection plane) and made it colinear to one of the lines (just to give myself a frame of reference.) Then I made a 2D shape & projected both lines & the curve onto the block. I made a copy of the resulting 2D shape & bisected the arc, creating 2 half shapes. Then I used the TriBall to twist one of the shapes from where the two halves of the arc met, until the line was parallel to the original line. You can see that they are not colinear & the arc has a different radius from your original arc. Finally, I created a 3D curve out of two line segments and two arcs.

 

If I knew which elements were critical, I could probably get it much closer - assuming that what you are asking for doesn't require contradictory geometric conditions (for instance, a line can't be tangent to an arc if it would cross the arc at two points.)

 

Here's my file - see if the concept works for you.

 

 

fullrad2.ics

 

[tlehnhaeuser]

Mike,

I think you got it. I will have to clear this with the client to make sure. This is high tolerance rebar to be used in a new concrete system and apparently the full radius is critical in their strength calculations.

Thank alot guys for all the help.

Tom

 

[cborer]

I like the idea with the cylinder cause it is a good sample for my old wish:

The 3d magnetic curve!

Here it would be nice to have the curve tangent on both lines and laying (magnetic) on the cylinder surface.

 

Or Kevins example:

Create a surface from the two lines (which will be 3d), extend it, and draw a 3d magnetic curve on this surface with both end tangend on the start lines.

 

But I think it wont be a part of a circle anyway, it will be a part of an ellipse.

 

Carlo

Edited July 14, 2006 by cborer

[Mike Allen]

I started looking at this from another angle & came up with another solution. This time, I kept the entry & exit angles of the lines & the exact radius of the arc (or the cylinder that the arc fits around.) I realized that a curve wrapping around a cylinder at an angle is going to be elliptical.

 

I measured the angles of the lines relative to the cylinder & to each other, then used surface reshaping to taper the end of the cylinder to match the first line. I arranged the original 3D curve so that the first line would be in the same plane as the tapered end of the cylinder. Then I copied the cylinder & tapered the other end, & arranged it so that it would match the other line as closely as possible (I had to fudge, since that line isn't tangent.) Then I created two 2D shapes - one that projects the first line and the top taper of the cylinder, the other that projects the second line and the bottom taper of the cylinder (on the second shape, I projected the line & then moved it so it would be tangent to the ellipse.)

 

The two projected ellipses don't intersect at a tangent point, but that's necessary if a single piece of rebar is going to wrap around the cylinder. So I moved one of the 2D shapes (and the corresponding cylinder, for reference) so that the two ellipses intersected as close as possible to a single point (this is where I really had to fudge, because the facets of the edges don't show the actual curve & I can't quite figure out how to find the point of intersection geometrically.)

 

Finally, I created a 3D curve by tracing over the two 2D shapes. I used two arcs - one from the end of the first line to the point of intersection of the ellipses - again this took several tries, because I couldn't get the intersection closer than a few thousandths of an inch - and a second arc from the end of the first arc to the end of the second line.

 

The arcs don't perfectly follow the cylinder, but they are extremely close & there is no 3D elliptical arc tool.

 

I also tried using the TriBall to place the points of the arc segments directly on the surface of the cylinder (hitting "p" for each point), the way you can with a spline (my workaround for Carlo's magnetic curve), but the arcs didn't want to cooperate. It's funny, because it works without the cylinder.

 

Anyway, here's my second attempt (actually, about the 50th - but the others were pretty unsuccessful.)

 

-Mike

 

fullrad3.ics

[Mike Allen]

I've been playing around with using the TriBall to position 3D curve points on the surface of a solid object & it does work with the 3D arc - but only when the "Create Smoothly Connected Curves" button is not depressed.