Full Version: Inconsistency in trim?

As part of my tests with MoI3D v5, I made an arc of 300 degrees and trimmed it with a circular array of lines, to divide it in evenly spaced sectors.
However for some odd reason, additional interpolation points appeared at a couple of positions.
I found that odd so I repeated the test with an arc of 270 degrees and no such points were added.
Why the inconsistency?

Attachments:
trimmed_arc.3dm

Image Attachments:
trimmed_arc.png 

---

If i make an Array / Curve of a Point with your file i have not your result! :)

For count the radial lines select them (right to left) and see Detaisl or the up Right Page Pannel
So here 51

---

If I make your Array / Radial Lines ....Ttrim with an arc of 300° (from your text )
I have the expected result trimming!

---

Yes while you are trimming it is shown correctly but the resulting trimed arc if you keep everything joined, has additional interpolation points in the case of 300 degrees which don't appear in the case of 270 degrees. That's what puzzled me.

I simply used one line and made an circular array of 60 elements, then deleted the ones I didn't need, by the way.

---

Hi geekmidget, it's from the automatic rebuild mechanism that shows replacement edit points on circles, arcs, and ellipses so that you will get a smooth result if you manipulate the points.

Exact circles are formed in NURBS by a sequence of arcs. The actual internal control points of your 300 degree arc are like this:



The problem with this is that if you manipulate one of those points the result will have a sharp corner in it like this:





So to avoid that when you turn on control points on an arc or circle you see the points of an automatically rebuilt curve that does not have interior corner points and so will deform smoothly.

Note that the location of the corner points corresponds to the spans that you noticed have 4 control points instead of 3:



This "auto rebuild" mechanism will use the actual control points if there are only 3 points in the segment. After trimming there are more than 3 points in the curve in those spots where the segments are glued together. That's what causes the different auto rebuild behavior.

- Michael

Image Attachments:
arc_edit_points1.png  arc_edit_points2.png  arc_edit_points3.png  arc_edit_points4.png 

---

Points of trimming are not Points of Control ... when you join several curves
Here there are moved slighly from the trimming points

You have now the explanation from Michael! :)

---

I understand this but in both cases (270 degrees and 300 degrees) there are control points falling somewhere between two lines, however the trim gets evenly interpolated in the case of 270 degrees and in the case of 300 degrees it inserts two points instead of one in those couple of places. I thought internally arcs were saved as mathematical representations.

Sorry I was trying to reply Michael but I don't know how to use quoting in this forum UI. Is it supposed to be done manually?

Image Attachments:
arc_270deg.png  arc_300deg.png 

---

OK So I made another related test, I used a circular array of points following an arc of 90 degrees (I added some markers for visualizing alignment), then used the freeform through points and the resulting 90 degree arc didn't respect points. So this isn't so much a trim issue as an interpolation issue.

Image Attachments:
90deg_arc_using_points.png  90deg_points.png 

---

As said previously you can't use Control Points as real Geometric Points!

---

I thought in the case I made a freeform curve using actual points they would be respected.

---

Hi geekmidget,

re:
> I thought in the case I made a freeform curve using actual points they would be respected.

It will be if you use Draw Curve > Freeform > "Control points".

In that case the points you pick directly become the NURBS control points of the resulting curve.

That won't be the case if you use "Through points", which calculates a curve that passes through the points you pick. The control points of the resulting curve will not be the same points that you picked, they will be the result of the interpolation calculation.

NURBS curves do not inherently pass through all their control points.

- Michael

---

I see, that naming seems counter-intuitive to me, but thanks! Now I get it.

---

HI geekmidget,

You seem to be trying to use the edit points as a diagnostic tool.

That's not really what they are intended to do.

In MOI the edit points are meant to provide a way to deform a curve. Since the regular NURBS control points for a precise circle or arc do not deform smoothly, there is a curve rebuild process applied to edit points. The rebuilt curve will be used if you move some of the points around.


re:
> I understand this but in both cases (270 degrees and 300 degrees) there are control points falling somewhere between two lines,

No that's not quite correct, the 300 degrees has "actual NURBS control points" like this:



And the 270 degree has this:


The 270 degree does not have any fully multiple knots interior to a trim span, they're all at the end of a span.

The 300 degree one has 2 spots where a fully multiple knot is internal to a trim instead of at the ends of a trim, those are the spots where the rebuild points will get 4 edit points.


> Sorry I was trying to reply Michael but I don't know how to use quoting in this forum
> UI. Is it supposed to be done manually?

Yes. You can put the quote inside <quote></quote> if you want.

- Michael

Image Attachments:
arc_edit_points5.png  arc_edit_points6.png 

---

Hi geekmidget,

re:
> I see, that naming seems counter-intuitive to me, but thanks! Now I get it.

Well, with the "control points" method the points you pick directly become the NURBS control points, I don't follow how that's counter-intuitive.

- Michael

---

Try to watch the UI from the perspective of someone relatively new to MoI3D who just sees the icons and names of tools, not realizing the internals.

Image Attachments:
freeform_method_icons.png 

---

Actually I was trying to get the curve to follow the points precisely in the plane XY first to then assign each a Z value.


Thank you!

---