Hi Marc,

> Maybe there could be another command?

Yeah, I guess since it would work pretty differently it would be a different command.

But the difficult part would be figuring out how it would work. I'm not familiar with any existing NURBS editing techniques that work in that way to reduce control point count just within a selection of existing control points.

Do you have any kind of an example that shows an input and the desired result?

If there are some kind of rules about how the desired result is generated that would then make it possible to implement it, but if the desired result is generated just by using your judgment, that is the kind of thing that is hard for an automated command to replicate.

- Michael

Hi nos,

>> It would help also if the controlpoints would be located
>> as similar to the control points of the input object as
>> possible. No idea if that would be possible.

>I can't see how that would be possible - the way that rebuild in "fit to tolerance" mode works is that it constructs a new curve that >tries to only follow the shape of the existing curve and not pay any attention to things like its current control point structure.

>It will put more points in areas where the shape has more local details like when it goes around a tighter bend and things like that.


hi michael

I think my question at the beginning was missleading, cause the basic problem is that the offset function produces a lot of points. So it's probably better to try to have lesser points at the beginning instead reduce them later. Many times I use the offset function I want to tweak the result. In this case it would be ideal if the control point of the offset-curve have a correlation to the source-curve. I don't no if this is possible in any case the offset function is used but it seams that the ofset curve an be generated by modifying the location of the control points.

Here's an example:



~nos

Hi nos,

> In this case it would be ideal if the control point of the
> offset-curve have a correlation to the source-curve. I don't
> no if this is possible in any case the offset function is used
> but it seams that the ofset curve an be generated by modifying
> the location of the control points.

No, it's not possible to perform an offset just by moving control points around, at least not and make an actual true offset curve that is a constant distance away from the original.

An offset function that only offset control points alone would instead make a curve that sort of generally resembled an offset but would not have the actual "constant distance" relationship.

Your best bet for simplifying the result from Offset is to run Rebuild on it with a tolerance value of something like 0.1 - that should simplify the control point structure of it quite a bit.

But after doing these kinds of various analytic and refitting processes on the curves you won't be maintaining the control point structure of the original curve, these processes work by looking at the geometric shape of the curve (like evaluation points marching along the curve) and not off of its control point structure.

- Michael

Hi nos, here's a quick example to show you that "control points offset" does not produce an exact offset curve, except in the special case of line segments.

Here I have made a polyline and then produced an offset of it:



Then I drew 2 control point curves (with Draw curve > Freeform > Control points) and snapped on to the end points there, to make these curves:



So you can see there just by eyeballing it that there is not a consistent thickness between the 2 curves.

That's because an offset operation on a line structure is not the same kind of thing as a scaling transformation - applying an offset on just a line structure changes the proportions between the various segments - some segments become longer and some become shorter. Because a line offset produces that kind of changing internal proportions it also follows that just offsetting the control point hull of a curve will also change the internal proportions of the curve.

- Michael

ah, I see, thanks for pointing that out.

~nos