Had a quick question for my geometry buddy. I have this file, seems very simple, and was wondering if I was missing something with regard to what I should know to acheive my results.

A guy in another forum asked for some help, and the package he needs help in has fairley rudimentary drawing/modeling tools/methods.

The shape seems simple enough, though while fooling around with it, noticed some G2 blends and conics seemd to fit well for the profile. There are some simple dimensions in the file, and I couldnt seem to find anything tangible to figure out "The exact shape" (If one exists!)


The shape will need to be drawn with "Point picking". The profile in question is this one shown here:



I understand a bit about the drawing of tangency and such Illustrated in the next picture with green arrows. The question is this...I created the shape with the given dimensions, though not exact in the specific curvature. Useing the blend and connic tools, I could also create "pretty darn close" the shape. Doing this and examining the points, I found these 3 tangent points with red arrows.



Is there something in these dimensions that would point to an "Obvious Conic" or that would describe to me this "angle for the top tangent points" to acheive an exact shape?

It may very well just be an arbitrary shape, in which case I'm over thinking it. Seems I can get stuck on the most simple curves like this with trying to figure out if it was formulated. ( I did this with a similar "double purchase dead eye shape" I was creating for a friend. Turned out the curve was just arbitrary and I was sure it was purposeful.)

Anyway, a quick look would be appreciated. Any help also appreciated.

Thanks

Hi Burr - my 2 cents is that the outline does not seem to be made up of any analytic curve like a conic or arc or ellipse or any predefined shape like that, it seems to be a more generic spline curve shape.

So you'd probably want to draw it with a control point curve and not with things like one piece a conic and another piece something else, because you will more likely be introducing some unwanted segmentation to the shape if you do that.

If it was explicitly called out to have segments like that, that would be one thing, but from what I see the intent is more to have a curvature continuous spline instead of any analytic segments.

- Michael

Thanks Michael,
Yeah, probably me just being overly mental. Seems if you drew a circle and turned on control points and did a drag of one point out to the side, I would get stuck analyzing the shape for "day's" trying to figure out "how it was created and what the intent was.

Thanks for looking at it.

Gee! thanks for putting me on the spot Burr ;)

Firstly, it's obvious there isn't enough information to tell how the form is made up, but we do come across this situation a lot at work when we receive geometry which is made up using a spline or series of splines and we do have to simplify it into analytical geometry (simple lines and arcs) for the CAM guys, or else they'll have a grizzle, so it is possible to break it down, you can't expect an exact match but we do get it within 0.0005" (0.01mm) sometimes closer.

Without the actual geometry I traced the picture as close as I could and show in the attached .3dm a series of trimmed tangential arcs that I used to make up the shape, you'll see off to one side (as shown in the jpeg) the arcs and there corresponding center points in the same colour as the arc, so if you pick on one of the arcs, MoI's information panel will show the radius.



Is this what you meant?

Cheers

Tricky arcs circle :)

>>>>Is this what you meant?>>>>

Well, Kindof. Your geometry shows "all Radiused Arcs" that makeup the shape "fairly close". "Turning on the control points for all your arcs will have them lineup as in the second picture I posted (with red arrows) where the 2 lined up. Since we are tracing a bitmap (not exact) I needed to look for some kind of "indicator" of drawing something "correctly" vs. Just drawing something "close".

Both wabble on and off the bitmap appearance a bit, but your model has some sort of "symmetry and order" vs my result which has no order. I would tend to think that the model with some sort of recognizable "order" to it's construction would be the original intent. ????

The basic question was more, "If something from this layout had you just say, "Look, it's a sphere and a triangle with a g2 fillet" or something very simple like that.

More was you just did a trace, and then your experience turned it into something tangible. Though I cant figure out how you got "all arcs" out of the shape.

Can you shed any light on this part?

Hi Burr,

I see what you mean, well....more so with 3d geometry, at first I did see it like Michael said as a generic spline curve shape, then I just wondered how many arcs it would take to make up the shape, but that's just me, as I'm tuned that way because of what I said about work in the last post.
Another point is, this is the only reference we have and we are reverse engineering the shape and so like digitizing a 3d shape you can only be so accurate and assumptions are involved too.
Sure...! but I did this quick so it could be out of symmetry.

Obviously the main tools are Circles and Arcs, defined by the '3 points' and 'Tangent' option.

Firstly I defined the front circle and the end circle using Circle>3 points. Remember, I'm just eyeing these off, there is no real accuracy,
the only accurate outcome of this will be the length, width and the fact it will be a continuous smooth shape.


Then it ended up needing two other arcs in between again using the 3 point method.
The final stage is to fill the gaps between these arcs with tangent arcs, some tidying up with trimming, and your done!


Let us know if you need anything else.

Cheers

Thanks Danny, Thats great. Probably the best input is that since the lack of information, Embellishment is needed, and even at higher end modeling (you) it is also!

I stumbled up on "analytics" thinking something was there, though I'm not algebra/geometry savy I cant see it.

I used the conic curve then broke it at it's tip, then ran a g2 blend on the wider end points and if you bulge it up a bit it can "look" exactly like this shape, though I could never find the points to use to start the conic that would have it end up at the size specified.

Anyway, At some point in life I would like to get some instruction on class A surfacing and become a higher end modeler...Maybe after the kids are well into school. Maybe wishful thinking.

Thanks again for the input :o
Burr

Does a simple arc circle (yellow) with moving verticaly "control points" is less legitim if curve is over the paper curve drawing?
(start arc / center middle was the middle of the total length )

I would construct from Ellipse.

>>>>Does a simple arc circle (yellow) with moving verticaly "control points" is less legitim >>>>>

No i dont think so at this point.

What I was looking for was if the curve itself was some sort of "perfect mathematical representation" that could be found by just looking at the "dimension figures". I didnt do well in triginometry/algebra/geometry in school and sometimes (though everything can be described with the math) simple things are very easy to spot by people experienced in these areas.

The package that this has to be re-created in does Not have "control points" with it's spline tools. So when drawing, it's either a freeform curve, or you need specific "coordinates or pick points". So, I can come close, by tracing the bitmap...But if there was some kind of recognizable structure to the curve, I could place "points" at specific coordinates, then just pick those with the spline tool and the curve would be exact.

So far the consensus is that it is "just a curve".

Thx for the precisions