wonky lofts in 2001+

Hello,

I am trying to make an aesthetically pleasing outer casing for a
client, and it's giving me a migraine. I have guide curves for the
side profile, and cross-sections (made from splines) for the levels to
loft over, but the surface is losing its tangency for unknown reasons.
I am doing half at a time, intending to mirror the surface later.
Since I couldn't control the orientation of the endpoints of my
splines, I added a straight line segment from the centerline for 3mm
before starting my spline, adding a tangent relation between the line
segment and spline. When I loft, however, the lofted surface does not
follow all of my profiles! On some (but not all!) of my profiles,
whereas my line segment is normal to my mirror plane, the lofted
surface is tangent, resulting in an unsightly ridge that is anything
but aesthetic. It later follows the spline part of the profile just
fine. I use the pierce constraint to make sure my profiles intersect
my guide curves.

Also, I have some extreme (for SW apparently) curvature that a single
loft can't handle, so I end up with 3 lofts, a sweep, and a fill
surface. These don't play well with each other or blend, resulting in
discontinuities in curvature after knitting, despite being derived from
the same sketches. And SW doesn't let me apply any kind of tangency
constraints between the surfaces.

On one of the lofts in particular, two of the three profiles are
straight lines, and the third is a spline. The resulting lofted
surface is nowhere near following the straight line, even after i broke
it up into as many line segments as i had in my spline. What is wrong
with SW (or what I'm doing with SW)? Why doesn't it follow the
profiles given it?

ProE can do what I want effortlessly with the boundary blend feature,
but my client uses SW, and needs to be able to adjust the case at a
later date, so I can't just import the ProE surface. Also, surfaces
made in SW seem to "wiggle" or undulate more than those made in ProE
from the same curves. No matter how carefully I make my splines, it
finds a way to buckle and not follow my profiles....

Please help me if you can!
Email me if you'd like me to send you the part in question.

Thank you.

 

Dude,

2001+ by this point is pretty antiquated. The best thing you could do
to help your customer would be to use a later version of SW. Lofts have
improved a lot in the past 5 years.

With complex shapes, it's very difficult to get help from a word
description. A picture really might help.

matt

 

hmm... how can i post a picture here? i don't have a web space to
upload them to.

thanks for answering so quickly tho. i don't know if i can get them to
invest in the new version.

 

Yes, I noticed the same type of misbehavior. SW still has a long way
to go with its lofts.

There was noticeable improvement in SW 2003 & SW 2004. Still problems,
though. I had one simple loft change spontaneously every time the part
regenerated. This was in SW 2005.

 

Try the loft without the guide curves, especially if you have plenty of
sections to define the shape. Adding guide curves causes all the
problems that you described in your opening description.

It sounds like you mixed up line segments with spline points. The
spline is a single segment, and if it isn't it likely ought to be. The
line ought to be a single segment.

Clamp out wiggles by breaking up into seperate lofts, using start and
end tangency.

Boy, you are using old software. Good luck
Ed

 

if i take out the guide curves, the surface is all over the place! i
thought i defined my profiles pretty well, and have them even more of
them closer together in the area of greater curvature, to try to help
it out a bit, but no.

my line was originally a single segment, but after reading about loft
problems when the different profiles had a different number of points,
i tried adding points to see if it would help. it didn't.

i did end up breaking them up into separate lofts, but it's not letting
me do anything about the tangency. that's when i get slight errors or
seams when i knit the results together.

so is 2006 the recommended package? is it stable? i know wildfire 1.0
was quickly superceded by 2.0.... but then, that was a big change from
2001

thank you all.

 

As Matt said, lofts have gotten quite a bit better since 2001+, but
tangencies are still sometimes an issue.

This is often not a good idea, because of tangency problems. I will often
build half a part and then mirror it, but if I run into problems at the
center line, then I go back to making the full part. It's a lot more
trouble, but you are more likely to get the shape you want.

This is pretty bizarre behavior. It is not unusual to have the loft not go
through the profiles. SW has a hard time figuring out how to follow the
guide curve and the profiles, so it elects to deviate from the profiles.

It's hard to remember what 2001+ was capable of. 2004 lets you have tangency
relations at the start and end of lofts, tangencies to the side surfaces of
a loft, tangencies at the edges of fill surfaces. I think 2006 gives C2
options for tangencies. Make sure you look at all of the options available.
You may be missing some possibilities.

Where you put your profiles and how you define them and the guide curves is
critical to getting the best output. Download Ed Eaton's tutorials from the
DiMonte Group website:
http://www.dimontegroup.com/

Work your way up from Curvy Stuff 101, through 102, 201 and 301, then Curvy
Stuff V Parts 1 and 2, with possibly a side route at some point through
Surfacing for Blockheads. You will learn a lot that will be very helpful
although the later tutorials may rely on features that you don't have
available in 2001+.

ProE is generally conceded to be better at surfacing than SW. That was
especially true back in 2001.


Jerry Steiger
Tripod Data Systems
"take the garbage out, dear"

 

As Ed implied, guide curves are a kind of sometimes necessary evil. If your
shape is an easy one for SW to do, you probably won't need any guide curves
at all. Guide curves take precedence over the profiles in any conflict
between them. The fact that your surface doesn't seem to follow your idea of
the expected shape without the guide curves may indicate that the placement
of your profiles is not very good.

Putting in more profiles seems like a good idea, but it usually is a very
bad idea. Getting the right location for the profile planes is usually the
right way to fix problems.

Having the same number of segments in your profiles will help make nicer
transitions between the profiles, but won't do anything for your tangency
problems.

I don't have 2001+ installed any more, but it seems to me that lofts always
allowed you to set start and end tangencies.

The surfacing tools generally get better every release. 2006 is a pretty big
improvement over 2004 (we skipped 2005). 2004 was a big improvement over
2001+. 2006 is fairly stable, at least as far as modeling goes. We have
problems with graphics (sketches that disappear) and crashing. (But we have
crashes in 2004 as well.) People have had problems with drawings and pdf
output and others that I have forgotten about, but it's not an utterly
hopeless release.


Jerry Steiger
Tripod Data Systems
"take the garbage out, dear"

 

I hear 07 has some good things in it for surfacing....

 

If you would like, I could host either an image of the part or the file
itself. I'm sure the advice would be much more constructive if you and
we were talking about the same things. As a warning, unless anyone here
still has SW01+ installed (where's TOP?), you won't be able to read any
SW native file anyone in a later version creates, other than a
parasolid. Email me.

matt(at)dezignstuff(dot)com.

 

http://mysite.verizon.net/mjlombard/wonky/wonky.html

here are the screen shots

 

All I see are four blanks with a little red x in each. The problem might be
on my end, as I have had problems with some other web sites. I've allowed
popups.

Jerry Steiger
Tripod Data Systems
"take the garbage out, dear"

 

your links are to local files (on your desktop), rather than networked
ones. just have to edit your links...

 

I can't see them - just red Xs in the corner of three frames.
Internet Exlorer 6.0.2800

 

woops.

learning a new html editor...

there, that should be better

 

thanks for the pointer to the tutorials! they were helpful. i moved
my profiles to occur at the points of my spline, and that helped the
surface a bit, but it's still not perfect. i'm trying breaking it up
into smaller chunks, and i'll keep you guys posted on my progress.
thanks for your suggestions in the meantime tho!

 

I see the images.
Oh man, there are so many things going wrong I don't even know where to
start.
Sorry - it would be a lot easier to just get a copy of the files and
fix it than to try to discuss on this forum, even if that meant
re-installing 2001+.
If you want more help, just post a valid email address and I'll tackle
it off line (the email that I use to post is no longer valid).

Othwerwise, good luck.

Some basic principles -
All nurbs faces are happiest when they are nice recatngles. Now, that
is stupid - we have to design what we have to design. But since you are
modeling in surfaces, you can break large shapes into smaller shapes
that better apporximate rectangles wihout wild contortions.
if you want to clamp out the wiggles, don't try to get a single surface
to do so much. The upper portion ought to be a single body, the lower
part a single body, and the transition a third body (it looks like you
tried to get the upper and transition out of a single loft which of
course creates issues)
the lower portion ought not have that crease - its either that your
sections allow it or there is a guide curve causing that.
To Jerry's point, you might need to chek your loft section spacing.
Basically - fewer sections = smoother.
Good luck - the basic principles are easy when you get them, but its
hard to bite off a bunch at once.

Ed

 

Thanks for the pictures, Matt.

One thing I didn't say earlier is that you don't need to add straight
segments in your profiles to get them to meet tangent at the center. You can
add straight construction lines at the ends of your profiles and add
relations to make them tangent to the profiles and perpendicular to your
center plane.

Listen to Ed's advice. (Something we all should do.) It looks like the top 3
or 4 profiles would make one fairly easy part, the bottom 5 or 6 another,
and the center is going to be a struggle no matter how you do it. I suspect
you will need to break the center up into a number of easier pieces. Make
the big flattish pieces that are perpendicular to your center line, then the
rounded piece that ties in easily to the ends, then use lofts or fills to
join the big flattish bits to the others. One of Ed's tutorials shows how to
build a bottle and that one would be especially good to follow carefully.

Jerry Steiger
Tripod Data Systems
"take the garbage out, dear"

 

Mark,

You know that Ed knows that!

I believe what Ed was trying to get across is that nurbs work best when the
shape of the patch is close to rectangular. As you distort the edges more
and more, the nurbs have a harder time following the geometry.

Jerry Steiger
Tripod Data Systems
"take the garbage out, dear"

 

Huh? The guy was asking about problems with lofts, not revolves and
extrusions.

Ohhhh, I get it. Good point - I should have made it clear that NURBS
faces are happiest when they have a four sided perimeter , each side
roughly perpendicular to the two sides adjacent to it (a ractangular
face in my mind is one with a roughly rectangular perimeter, whether
that is purely recatangular or if sides bulge out, squish together a
little, or whatever. The parallel is thinking about a triangular face
which can be lofted but often idn't happy about it, or N sided faces
which require surface fill or some tricky predcition and trimming to
achieve with lofts).

I can see how the sentence I wrote could have been read that I was
talking about planar, rectangular faces, which would have been no
f***ing help to the guy at all. Boy, this jsut goes to show that its
even more complicated to talk about this junk than I alluded to in my
post.

PS - Jerry, thanks for getting my back

Ed