Extrude Along Path - why's my wall leaning over?

[line-weight 474]

Both of the EAPs in the attached file are supposed to be walls, which follow a 3d path (curving in plan and changing in elevation along their length).

 

I have 'lock profile plane' ticked. My understanding was that this means that the profile remains perpendicular to the X-Y plane as it follows the path (thus creating a vertical wall that doesn't lean over). This is not what's happening though.

 

Am I doing something wrong?

 

Profile:

 

 

resulting EAP viewed in elevation:

 

 

VW2018 file:

 

Untitled 23.vwx

 

 

Untitled 23.vwx

[Benson Shaw 405]

It seems that the Lock Profile Plane is supposed to control the profile orientation so that it sweeps always normal to the ground plane. But I have found that it doesn't control in that manner.  EAPs in this file are made from paths which curve through several planes causing a twist - ie Normals along the path change direction.  I THINK the EAP lock establishes a sort of localized x1,y1 plane for each position along the path causing the twist and lean in the extrusion.

 

I don't have a great approach for this. Only a couple workarounds.

 

A simple, vertical, rectangular section can be made with a loft and a shell: duplicate the path,  raise the dupe to top of wall height, Loft no rail the two curves thereby creating a vertical, curved "sheet"), Shell to wall thickness.  The little projection can be approximated: Extract the top edge of the wall, move extraction down to top of projection, duplicate, move to bottom of projection (that makes 2 NURBS Curves laying on surface of the wall), loft no rail the new curves, Shell (opposite direction from wall shell), Add Solids.

 

Or

 

Duplicate the curve, move (z) the dupe to top of wall height, Convert wall profile to NURBS or 3d poly, snap it between the curves at one end, rotate in top view to be perpendicular to the curves, Loft Birail Sweep.  Problem with this is that the sweep does not stay tangent/perp to the curves, so the solid sweep will flatten at the sharper portions of the curve.

 

sigh

 

-B

Edited October 9 by Benson Shaw

[line-weight 474]

Thank you for your reply @Benson Shaw and your suggested workarounds.

 

Not sure if this counts as a bug? Anyone from VW want to comment?

 

This is not unrelated to my other thread (linked below) where I am pointing out that the documentation of EAP is insufficient.

 

And this is such a familiar scenario in VW - I've not used EAP that much in the past, but thought I'd try and get my head around it. So I invest a load of time trying to understand how it works, and the answer is that it doesn't work properly. Like so many other things in VW.

 

As you say - sigh.

 

 

 

 

[line-weight 474]

Ok, after a certain amount of testing, and also following some clues I found in this useful thread on a similar subject:

 

https://forum.vectorworks.net/index.php?/topic/52320-extrude-along-path-rotation/

 

I think I have found how to make it do what I want.

 

1) The profile must be a 3d object (3d polygon or NURBS curve seem to work)

2) The profile must be oriented in 3d space, as you want it. So it must be placed so that anything you want vertical is vertical. And it must be placed how you want it relative to the normal at the start point of your path viewed in plan (ie its rotation in plan is relevant)

3) When you preform the EAP command, "Lock profile plane" and "fix profile" must both be ticked

4) Now if you're lucky it'll produce what you want

5) Now (and I think this is a result of ticking "fix profile") when you edit the profile, instead of being taken to a top/plan view of the profile you are taken to a top/plan view of the path with the path greyed out and the profile also viewed from the top. The origin of this view seems to be the start point of the path, and you can move and rotate the profile relative to this. To edit the profile, you have to edit it as a 3d object. In my case, to alter the profile, this meant setting a working plane on the plane of the profile, then looking at this plane.

 

Here is an example file

 

Untitled 1.vwx

 

Here's the EAP viewed in plan

 

 

and in elevation

 

 

Having thinking that I've worked this out, some problem will probably arise now that I try and use it for real, but let's see...

 

[line-weight 474]

*FRANTIC WAVING AT VW STAFF*

 

Is anyone from Vectorworks reading this?

 

The EAP command needs proper documentation. If this was all explained properly in the documentation it would save me and others spending literally hours working stuff out by trial and error when there should be no need to do so.

 

This was already requested in a thread in 2017, two years ago, the one I link to in the post above. As well as the one I posted a couple of days ago which no-one from VW has responded to.

[Benson Shaw 405]

Yes, Doh!  Good sleuthing @line-weight . Seldom used, so I have to relearn: 3d profile, orient and snap to end of path, Lock and Fix.  I think for such a wall like object rising vertical relative to the ground plane, no need to establish a special working plane. I watched the help video just now.  It claims that the Lock Plane will preserve the verticality. I think this example disproves that claim. Maybe because the path spirals through 3 space?  Video does not elaborate on the Fix and Lock combo which is needed here.

 

Another problem in this scenario is how to orient the profile so that its top view is perpendicular to end of path. Even a few degrees off causes the EAP to lose thickness.  Snaps with Rotate tool do not report the 3d normals of the path.  I use a 2d guide line to approximate.  In top view, zoom to end of path. Snap a short line in center mode to end of path and end it on the path a short distance from end of path.  Rotate 90° (Cmd L on MacOS).  Shorter line is closer to perpendicular. Rotate the profile with ctr and end snaps on this guide line.

 

-B

[line-weight 474]

5 minutes ago, Benson Shaw said:

I think for such a wall like object rising vertical relative to the ground plane, no need to establish a special working plane.

 

When I mentioned the special working plane, that's only if you want to edit the profile retrospectively (ie after you've done the EAP).

 

 

5 minutes ago, Benson Shaw said:

Another problem in this scenario is how to orient the profile so that its top view is perpendicular to end of path. Even a few degrees off causes the EAP to lose thickness.  Snaps with Rotate tool do not report the 3d normals of the path.

 

Yes, I noted this same problem. Could be troublesome if you need to do something very precise.

[Wes Gardner 169]

@line-weight,Mine's not leaning over...I used a polygonal object and a NURBS curve as the path with Lock Profile Plane ON.  I also noticed that your objects in the Untitled 23 file are about two hundred and ten MILLION units from the origin????

 

Wes

 

 

[line-weight 474]

16 hours ago, Wes Gardner said:

@line-weight,Mine's not leaning over...I used a polygonal object and a NURBS curve as the path with Lock Profile Plane ON.  I also noticed that your objects in the Untitled 23 file are about two hundred and ten MILLION units from the origin????

 

Wes

 

 

 

Did you use a 3d polygon, and what plane did you draw it on?

[line-weight 474]

16 hours ago, Wes Gardner said:

I also noticed that your objects in the Untitled 23 file are about two hundred and ten MILLION units from the origin????

 

 

P.S. well spotted - this is also true in the working model I pulled it from. I don't think it's the cause of the problems but I have now fixed this in the original model to avoid other problems cropping up. Thanks.

[line-weight 474]

On 10/9/2019 at 9:48 PM, Benson Shaw said:

Yes, Doh!  Good sleuthing @line-weight . Seldom used, so I have to relearn: 3d profile, orient and snap to end of path, Lock and Fix. 

 

Coming back to this ... have realised, it's important to get the relationship of the profile and path correct in the Z direction when you create the EAP.

 

The reason for this is that going to 'edit profile' lets you accurately orient the profile relative to the path on the X/Y axes but there seems to be no indication in this view of where it is relative to the path in the Z direction.

 

It would be much more useful to have an 'edit' view where both the path and profile were visible (and editable) in 3d at the same time.

Edited October 14 by line-weight

[line-weight 474]

On 10/9/2019 at 6:41 PM, line-weight said:

 

 

 

Having thinking that I've worked this out, some problem will probably arise now that I try and use it for real, but let's see...

 

 

 

Guess what - it's Vectorworks so of course some other problem arose.

 

I had some success building a number of viaducts which are curved in plan and rise/descend along their length.

 

 

The problem is with the ones with more gradual curves, because VW seems unable to render these without introducing very large facets, making them look awful, and making it impossible to align/transition with other objects accurately. Discussed in another thread (  https://forum.vectorworks.net/index.php?/topic/55522-smoothing-nurbs-curves/  )

 

A potential solution to this is to change the NURBS path so that it is segmented with sharp corners, making the segments as small as necessary to give the appearance of a smooth curve. However it turns out that doing this causes the EAP to become hollow, meaning I can't subtract from it. I think that's a bug, if anyone cares.

 

Edited November 28 by line-weight