Is there any chance of creating a Factory to find the Envelope of multiple copies of a curve (2D), or a solid (3D)?

Apparently it would require the derivatives, or partial derivatives, of NURBS curves. With some kind of solver...
And be very hard to do...

- Brian

A blob ? Metaballs...

Thank you Pilou.

Maybe envelopes of multiple arcs, lines, and spheres is a little bit easier?

- Brian

So with all geometry in Top View, an arc or a circle, with center point selected, will Sweep a rail curve, (of large enough radii), SOMETIMES produces an envelope.
More testing needed...

Sometimes an "envelope" is produced, sometimes a 3D sweep occurs.
Placing the start of the rail curve at the center of the circle yields 2d.
Then moving the rail curve causes change to 3D..


Maybe I can use this method for a project, or not.

- Brian

I am unsure when "Auto Placement" of the profile occurs.
Can there be a version of sweep with "Auto Placement" turned off?

Sweeping a circle or arc with its center selected...
Sweeping a circle or arc, with a point on the arc, or off of the arc, but on the unused extension of the arc...

Iss Auto Placement and Auto orientation turned off in some situations?

- Brian

Hi Brian,

re:
> Is there any chance of creating a Factory to find the Envelope of multiple copies of a
> curve (2D), or a solid (3D)?

Sorry no there isn't any functionality in MoI currently for calculating this.


> Iss Auto Placement and Auto orientation turned off in some situations?

It's turned on if the profile curve is outside of the rail curve's bounding box.

I'd like to change it at some point to be able to be manually controlled but there isn't an option for that currently.

- Michael

Thank you very much Michael.

Bounding box is interesting information.

I have been having fun, after a circle sweep to a curve, moving the curve around the screen. [Or moving the circle around the screen...]
The 3D sweep follows along, (due to History I think), and changes to planar "envelope-ish" when part of the curve is in the (it turns out), Bounding Box of the circle.
(or is it when the circle is in the rail's bounding box...?) I am not sure... The curves bounding box seems to jump in size...?

The 2D "envelope-ish seems to have a bit extra at the beginning...
Including the center or point on the curve seems to have some effect on the sweep???

I'll have to read up on Bounding Box.

- Brian

So selecting multiple objects expands the (combined) boundary box to a rectangle, encompasing all of the objects.
Each object still has its own bounding box...

History is ended if the sweep result only is moved...

Extrude > Set Path makes what looks like a sweep surface.
The boundary should be the envelope,
But it looks like the object is "degenerate". It does not seem to have an outer boundary.

- Brian

Hi Brian, Extrude with Set Path will make what is called a "Sum surface". It's a surface made from 2 curves where all the U direction isoparms will be from the first curve and all the V direction isoparms will be exactly the shape of the second curve.

An extrusion is one example of this type of surface with a line being one of the curves.

I think that I didn't understand properly what you were looking for, I kind of thought you were looking for something like the convex hull around a set of curves or surfaces.

Does the Offset command applied to your path curve give you the side pieces that you need?

Trying to generate a sum surface using Extrude with Set path as in your example will not be good because the result is a self intersecting surface that folds back over on top of itself.

- Michael

Hi Michael,
Thank you for the explanations.
I've been experimenting/messing around, trying to manage an "envelope" of a family of one 2D profile, offset in the same 2D plane, along a path.
My profile will be composed of 4 circular arcs, but the tests were with a circle, for simplicity.
This is sort of like a convex hull...
So Extrude does not create (in this case) a useable desired result for the circle.
Planar sweep of a planar circle does not create (in this case) a useable desired result.
The 3D tube created by sweep, is promising for a circle, but using part of a circle gives variable results, due to the auto placement I guess.

( As in other posts, a few hundred boolean differences is not practical...:-) )

I'll post my 2D object, after it gets created, in a few days...

In theory, the parametric equations can be solved with Matlab and calculus, likely beyond my ability.

- Brian