Circular Loft?

[ulrich1a]

There were recently to help requests in the forum, where a circular loft would be a solution.
viewtopic.php?f=3&t=6143
viewtopic.php?f=3&t=6127
I understand under a circular loft, a loft were the starting and the end profile are the same. Is it possible to add such a feature? Like an option you can check: use start profile also as end profile.
Allows OCC such an feature?

Ulrich

[jmaustpc]

If you use Part Sweep rather than Part Loft ...it works.

here is a file with three examples including one similar to the one from the other day with the circles tilted around a circular path.

circularsweep.fcstd

(32.99 KiB) Downloaded 93 times

Of course whether or not Loft should also work is another matter.

[NormandC]

Closing a loft is possible in other software so I think it could be useful, if it's technically feasible.

[jmaustpc]

Closing a loft is possible in other software so I think it could be useful, if it's technically feasible.

I made a feature request on the bug tracker

http://www.freecadweb.org/tracker/view.php?id=1498

[wandererfan]

I've started looking into issue #1490, but I haven't used loft much, so need some guidance.

Attached file CircleLoft2.fcstd looks like it makes a "circular" loft. It is a trial hack made by duplicating the first profile.
a) Is the result (Loft002)supposed to look "lumpy" like it does? I guess I kind of expected a smooth torus.
b) Loft002 has errors in Part.CheckGeometry, and they don't go away after "Part.RefineGeometry". I guessing a certain amount of self-intersection is inevitable.

From a scripting standpoint, we want to go from Part.makeLoft(profileList,isSolid,isRuled) to Part.makeLoft(profileList,isSolid,isRuled,isClosed)?
From Gui standpoint, we want a "Closed" check box in the task panel?

tks,
wf

[quick61]

Hi WF,

Is the result (Loft002)supposed to look "lumpy" like it does? I guess I kind of expected a smooth torus.

In your Edit > Preferences > Part design > Shape view tab your Tessellation setting looks to be at the default 0.5. If you change that value to something like 0.02 (where mine is set) you will have a much better looking model. This does have it's tradeoff as your more complex Boolean operations, producing Spheres and the like, will be a bit slower. I'v found that the 0.02 setting is a good middle point between performance and smooth surfaces. Using the latest OCC/OCE has some advantages in this area as well.

Loft002 has errors in Part.CheckGeometry, and they don't go away after "Part.RefineGeometry". I guessing a certain amount of self-intersection is inevitable.

That's the rub, how to get rid of the self-intersection. On this I have no clue. Sense we can rotate a sketch and have it come out without a self intersect, one would think it should be possible to do with a loft, being able to use the starting sketch as the end point. Maybe look at the rotate function in Part Design? Probably not it but...

Mark

[ulrich1a]

From a scripting standpoint, we want to go from Part.makeLoft(profileList,isSolid,isRuled) to Part.makeLoft(profileList,isSolid,isRuled,isClosed)?
From Gui standpoint, we want a "Closed" check box in the task panel?

Yes this is, what I thought of. I would also expect a round loft then and not this nice heartly formed.
When I tried to make a closed loft in a macro, I got additional surfaces at the doubled profile and a crash, when I put the same profile at the start and the end in the list.
It is interessting, that you got a closed shape.

Ulrich

[jmaustpc]

I would also expect a round loft then and not this nice heartly formed.

I think this is an example of where you guys are asking to much of Loft. If you want a round, or anything else, path then I think you should use sweep and define the path or if more appropriate one of these other tools.

We have
Loft,
Sweep,
Revolve,
Extrude,

which are all similar to Loft in some ways.

@ WF
I had a look at your CircularLoft2 file, great example by the way, It demonstrates exactly the behaviour of Loft that I would have predicted. Both in the shape produced and the way the path through first/last profile is transitioned. That transition is exactly the same problem we had with the Bezier curve (during its development) transition when it was switched to "closed". Perhaps that similarity might give you some ideas on how to code this matter?

The best way to think of this is to compare it to the Sweep. The main difference between the sweep and loft is that with the sweep you define the path. So a Loft from a user's perspective is just a sweep with an automatically allocated path based on one of two transition/path modes.

The default currently with "Ruled False" uses what appears to me to just be a Bspline line path, like Draft Bspline, where the path is translated through a series of points defined by the profiles. If the Ruled parameter is toggled to True, then the effect is what you would expect from a Sweep using a Draft Wire as the path through those same points. In the True case the transition is predicable and the close is a "no brainer" but with the default "false" you have to recalculate the whole path if it is closed. Otherwise you get what you have shown in your example where the transition is not a continuous Bspline, it rather is like a lineal transition.

Here is some screen shots to show what I am talking about.
This is the default as is in your file which is an example of a loft being the same as a Sweep with an auto calculated Bslpine "open" path, open in the sense that even though it starts/finishes at the same spot, the transition is not a continuous Bspline, its rather a stop and start, not another through point.

loft_ruled_false.jpeg (115.74 KiB) Viewed 5365 times

This is what it looks like with ruled "true", In this case it is similar to the result expected from a Sweep with a wire or straight line segments for a path.

loft_ruled_true.jpeg (108.62 KiB) Viewed 5365 times

The effect of the transition through the first/last profile is more obvious if you move your sketch001 down in the Z axis, thus giving an equivalent to a Sweep who's path is a 3d Bspline. Like this

loft_ruled_false_sketch001_down_in_Z.jpeg (110.62 KiB) Viewed 5365 times