Full Version: Sweep fails - why?

Show messages: All  1-4  5-20

Use the OrderCurves script to view and change the direction of the vectors.

OrderCurves
https://moi3d.com/forum/index.php?webtag=MOI&msg=9781.5
https://moi3d.com/forum/index.php?webtag=MOI&msg=9781.13

Wayne

Image Attachments:
Screenshot 2025-10-15 094344.png 

---

Hi Wayne, thank you very much ***
Mala
P.S.: Hui, WOW...that is so perfect...(also for my design!)

---

Hmmm´...I was a little too happy. It works with one geometry, but not with this one. Any ideas? Basically, the geometry is almost the same. The only difference is that the upright bar is bent slightly to the left...

Attachments:
sweep fail 3.3dm

Image Attachments:
sweep fail 3.png  sweep fail 3_.png  sweep fail 3__.png 

---

Try moving one pair of arcs to the opposite side of the centre curve.
As it is now they are on the 'wrong' side for it to work without twisting.

Martin.

---

And your 4 sections are not align on perpendicular planes! (first post)



PS 2 rails are more prudent!

Else why not extrude sections from the good green? ;)

---

Hi.
to make this work sweep first then in top view rotate the 2 left hand curve sections
180 degrees this will update the sweep to give the desired result.
Cheers
Barry

---

Hi Barry-H,
thx, but I need the profiles like/where they are.

---

thx, Pilou...because the GoodGreenGeometrie is a variation. I also need the red variation. It is the same, but if you look from front the vertical line points more in the left direction.

---

Martin, thx for your suggestion. I need the arcs point their directions like it is now.

---

Here's a relatively simple geometry. Here, too, the sweep is twisted. As workaround I wasn't able to create a double spline as Frenchy suggested. I think I'm doing something fundamentally wrong.

Attachments:
SweepTesting Fail 4.3dm

Image Attachments:
sweep fail 4.png 

---

Hi Mala,

re:
> Here's a relatively simple geometry. Here, too, the sweep is twisted.

If you make a sweep with just one profile at the start, it will reduce twisting in the generated sweep, like this:





Compare the end with "natural" twisting with the 2nd profile that you are using:



Basically your 2nd profile is at a pretty significant rotation away from the natural sweep result.

When there is a fairly large rotation from the "natural" position it can make things difficult for the mechanism that determines if a profile needs to have its direction flipped and also which segments should be connected with each other.

One way to solve it would be to add an additional profile in the middle. When there is a smaller rotation happening between each profile station it will help the automatic seam and flip detector work better.

Another way which is probably easier though would be to only do a sweep between a single profile segment instead which will eliminate segment matching and you can fix the wrong direction by using the "flip" command after the sweep is done.

So for your case here, select your profile curves:



Run Edit > Separate to break it apart into individual segments:



Select 2 of the segments that you want to be matched:



Run Sweep on those:



Finish the sweep and then correct the wrong direction by selecting the 2nd profile and then run the Flip command - type Tab then type Flip and push Enter. That should get this result:



Repeat this with the other profile segment (the line), then select the 2 pieces and use Edit > Join, then select the object and run Construct > Planar to put end caps on.

Result .3dm file is attached.

- Michael

Attachments:
SweepTesting fix.3dm

Image Attachments:
mala_sweep_twist1.png  mala_sweep_twist2.png  mala_sweep_twist3.png  mala_sweep_twist4.png  mala_sweep_twist5.png  mala_sweep_twist6.png  mala_sweep_twist7.png  mala_sweep_twist8.png 

---

Hi Mala, also the same steps will also work for your sweep fail 3.3dm file above.

result attached.

- Michael

Attachments:
sweep fail 3 fixed.3dm

---

Maybe more simple...
And advantage you can have start and target section different... here 2 circles but...

Make a boolean Diff between cylinder and plane

---

Ok. I learned very much (again). Thanks to you Michael and also for the smart way, Pilou!!! Now I can progress...fine.

---

Hi,
attached photo shows a method to get the twist direction you require. By creating a surface with desired twist.
Then array required section along curve aligned to surface and finally loft sections.
Cheers
Barry

Image Attachments:
Twist Orientate.png 

---

Thank you!

---

Show messages: All  1-4  5-20