Michael

I've been getting this problem since V1 I think.

Whenever I create 2 rail sweeps similar to attached with a scaling rail curve I get folding and creasing at one end with lots of points bunching up there.
It this case I get the bunching at one end whether using maintain height or maintain tangency.

Is it just bad modelling or a bug ?

Hi Kevin,

First of all, with your curve setup Network will achieve the same result and no bunching at the end.

If using Sweep the scaling rail should cover the full extent of the sweep, in your case the scaling rail is just short of the end of the rails that's why you're getting a bunching effect at that end because sweep gets to the end of the rails then has to come back on it self to honour the scaling rail length.

If you extend the scaling rail past the end of the sweep rails then sweep and you will be happy :)



Cheers

It might be to do with the Scaling Rail intersecting the large profile at an angle (in Front View). I was able to get a clean Sweep :-) by extending the Scaling Rail past the large profile up to the surface in the attached file.

 ***Beaten by Danny***

Martin.

Thanks Danny,

The only problem I have with this solution of extending the "scaling rail" past the "end profile" is it is kind of tricky to setup so that it intersects my end profile exactly.

Yep, your probably right, the sweep with scaling rail option is probably not something you would ever need as you would just use network in most instances although the maintain tangency option comes in handy in some circumstances which network doesn't have.

Just had a look at the command reference and in there Michael shows the scaling rail extending past the end profiles.

Hi Kevin,

> The only problem I have with this solution of extending
> the "scaling rail" past the "end profile" is it is kind of
> tricky to setup so that it intersects my end profile exactly.

You can just extend what you have to a surface like Martin explains in his attached file.

-

>You can just extend what you have to a surface

Yep, that's the only way to extend a line that I know of when you can't easily align the cplane to the curve.

Hi Kevin, yeah that sort of thing means the scaling rail is not being hit and so it's not extending quite far enough. When the sweep is not able to hit the scaling rail at a particular point, it reverts to the original unscaled sweep at that spot and that kind of sudden jump in shaping makes that kind of result that you saw.

The way the scaling rail works is the thing that needs to hit it is the plane of the "sweep frame" at each spot along the rails which is not necessarily the same thing as where profile itself actually is at, especially if the profiles are at some angle to the rails.

In the future maybe it would be good for me to internally extend the scaling rail if the sweep runs into this type of thing where it isn't able to find an intersection...

- Michael

Ok, thanks for the explanation

For some reason snapping doesn't seem to work perfectly when dealing with these curvy shapes, it always seems to miss by the tiniest amount and you have to do max zooming to find the error and even then it is difficult to get it to snap to the curve. Possible reducing the snapping radius in preferences might solve this problem, it'll be something I need to experiment more with.

>In the future maybe it would be good for me to internally extend the scaling rail if the sweep runs into this type of thing where it isn't able to find an intersection...
Yep, Anything that makes it easier for users is welcomed. Perhaps a curve extend tool also might solve it where the user would click the end of the curve and it would extend by the input amount. Anyway something for V4 or beyond.

Hi Kevin,

> For some reason snapping doesn't seem to work perfectly when dealing with these curvy shapes,
> it always seems to miss by the tiniest amount and you have to do max zooming to find the error
> and even then it is difficult to get it to snap to the curve.

The next time you run into this please save it off and send it on over to me so I can try to see what's going on there.


> Perhaps a curve extend tool also might solve it where the user would click the end of the curve and it would
> extend by the input amount. Anyway something for V4 or beyond.

Yes definitely some more ways for curve extensions are needed. I hope to be able to get to some of that in v4.

The tricky part tends to be that I try really hard not to have a whole bunch of separate commands so I'd be trying to work it into the current extend tool rather than having a bunch of different extend commands.

- Michael

>The next time you run into this please save it off and send it on over to me so I can try to see what's going on there.

See my original post for a file. You can see the error on the scaling rail when zoom in at the big end. All osnaps were on when drawing the scaling rail curve.

For the scaling rail I clicked on the top of the big end curve with a "on" osnap in perspective view, I then went to the side view and plotted the other two points and then I went back to the perspective view with another "on" osnap on the small end curve. Originally it missed both times on each end curve but mostly it misses on one of the end curves.

Perhaps I need to zoom in more when drawing my curves and the "on" osnap would work better ?

Sorry I can't post a video to explain how I did it but hopefully my text make sense.

Hi Kevin,

> See my original post for a file. You can see the error on the scaling rail when zoom in
> at the big end. All osnaps were on when drawing the scaling rail curve.

When I examined your file, I do see an error in the sweep surface on the big end, but I can't really see any osnap error in the scaling rail not touching the end profile there, it seems to end right on it... But like I wrote above it's not necessarily the _profile_ that the scaling rail has to cover, it's a set of planes that come up from the rail curves that the scaling rail has to cover.


> Perhaps I need to zoom in more when drawing my curves and the "on" osnap would work better ?

It can be a bit easier to control things in general when zooming in, but it shouldn't make any difference for whether "on" snap is on a curve or not - when you get on snap the point should definitely be on a curve...

Could you please give some more info on the snap error, right now I'm not understanding it.

- Michael

Hi Kevin, so something else that I don't quite follow:

> For the scaling rail I clicked on the top of the big end curve with a "on" osnap in perspective view, I then
> went to the side view and plotted the other two points

So with this sequence, it should result in those first 3 points of the curve all being on the same plane that's parallel to the side view.... But yours are not..

- Michael

Hi Michael

My mistake, the error is on the small end.

I think when I drew this I must have used the side view to snap to that last point. Because the small end curve isn't exactly perpendicular to the view it finds the nearest "on" which ends up slightly off.

When using perspective view to draw the first and last points of the scaling rail, all is working well. I cannot recreate any errors.

Sorry for the red herring.

Hi Kevin, no problem, let me know if you run into another kind of problem case in the future.

You wrote:
> I think when I drew this I must have used the side view to snap to that last point.

Yeah that can cause this kind of result because "on" snap (same as any object snaps) in the 2D views like the side view will get projected to the current drawing plane. That's so that when you're drawing things in a 2D view that you'll also get a planar result.

You can change that behavior with a setting under Options > Snaps > Object snap options > "Project to plane in ortho views" checkbox. When you unset that then object snaps will go directly to the 3D point snap location with no projection. But that also means that things that you draw in 2D viewports can be easily non planar curves squiggling around in 3D.

By default that projection is enabled so that in 2D viewports there's an emphasis on generating 2D results but picking points in the 3D viewport lets you do freely squiggling stuff.

- Michael

Thanks for the response, I much prefer projection being enabled without having to setup a workplane or cplane. Its a much more intuitive way to draw things in 3d.

Just got to remember that there is a disadvantage to this functionality when you work in the ortho views.

Yeah I've run into that problem w/the sweep and didn't understand why the bunches at the end were happening as well!


It would be great, Michael if you could internally make the curve extension - it's helps avoid breaking the "modeling zen" where "stuff just works" as expected.


-Will