Best Practice for Sweep

[Berserk]

Hello,

Is there a best practice for sweep?

I am currently working on a couple of window garnish.
The section has to be a certain distance to the steel frame.
I have created sketches normal to the steel frame and create a couple of sweep along the main lengths.

Then I use bridge curves and through curve meshes for the corners, sew them into a solid, then unite all of the sections.

I noticed that on some of the sweep, there are certain areas that looks slightly "crumpled".

I fixed the one on the front by changing the section to become bigger, so when I do a replace face, it actually remove material instead of adding. This fixed the surface that wasn't smooth.

On the rear sweep, there are crumpled areas and what I did to fix it is to remove one of the "guides".

I am just wondering if there is a better documentation or some guidelines with regards to sweep.



UGNX5.0.4.1 MP6 \ WinXP-SP3
Productive Design Services

http://www.productivedesign.com

 

[jerry1423]

Can you show us a picture?

 

[Berserk]

Here is a jpg showing the rear area

UGNX5.0.4.1 MP6 \ WinXP-SP3
Productive Design Services
www.productivedesign.com

 

[Berserk]

Here is the one for the front

UGNX5.0.4.1 MP6 \ WinXP-SP3
Productive Design Services
www.productivedesign.com

 

[jerry1423]

As far as being crmbled, you may want to open up the tolerance to see if that helps. I assume that you will be able to do that when you edit the sweep.

 

[hudson888]

I don't know if you're kidding yourself that those zebra stripes are going to be controllable using sweeps but if I wanted a nice shape rather that simply and engineered one then this is what I'd do...

Firstly I would have as many guide curves as I could or needed to to control the section shape and orientation. You can get up to four if you use a spine curve. For pure accuracy that makes all the difference. Secondly I'd situate my sections at the ends of the sections of the spine curves. If you have a straight section that meets a change of direction in a corner with less than g2 continuity then the system may approximate to produce an OG in the surface, better to break up your swept sections. But wait now you have a problem because sweeps have no way of matching end to end so you may have to build the corners later with curve mesh or even studio surfaces.

Mind you theoretically with guides build to great continuity a series of sweeps will match end to end but keep checking those edge conditions to make sure. I come back to the same point at the end when I say that the curve you use are all important. In fact if the curves you used were good enough to end match successive sweeps then you could probably only do better to sweep them all in one hit.

Taking a step back; I'd also look at your sections as four sided so what you really want to build are either four separate sweeps, one for each side, or one good one using preserve shape. Using preserve shape requires that the alignment method is parameter rather than arclength. Here you need to have nice even controls within all of your guide curves or you may risk wine-glassing within the individual faces. Turning on your UV curve display may help you verify this, but the real point is how you control the surface internals of whatever you build. Using arclength between two guides generally spans between them evenly, but with a closed section in a sweep like yours the section may be approximated to one or more likely two faces rather than four. It may look to have sharp corners when shaded but they're not. I hate this behaviour so its all I can do to warn you how to avoid it because once you create such a surface trying to work with it is usually far more trouble than it was worth.

After all that if you've well built guide curves with even parameters and good continuity through the corners matched with properly oriented sections then turn on parameter alignment and preserve shape and you ought to be able to create great sweeps that work every bit as well as you might expect. It is usually worth the effort of cleaning up your curve construction as opposed to manually struggling to eke out face by face one surface at a time. The problem is that nobody tells you how its all in the curve construction!

Now if you want to actually twist the visible surface through those corners to control their highlights then you're up against a creature of a different kind. The best advice there is that you will need to have guides that can be manipulated to produce the desired result, or you need to build the corner with more controls than the sweep can provide using other surface building methods.

There isn't really a best practice for sweeps per se, they're just another type of surface creation tool, so the real standard you need to apply is whatever works based on sound surfacing principles and understood standards of surface quality in order to produce a good result, often by trial and error.

Best Regards

Hudson

www.jamb.com.au

Nil Desperandum illegitimi non carborundum