symetry with lofts

Hi,

I do a lot of surface lofting, and following the advice of people like Ed
Eaton I'm trying to use splines to create ever smoother curvey shapes. One
thing I am finding is that it's really hard to ensure symetry and tangency
along something like a handle, where you can either loft one side and then
mirror, or use a closed loft and do the whole thing in one.

Has anyone got any tips on ensuring tangency and symetry. Really I want a
symetry plane option in the loft dialogue, so that I can ensure symetry and
tangency all in one swift move!

If I mirror and use Right-click>select tangency on one face most of the time
it doesn't see the mirrored face as tangent. However, the deviation analysis
says the devition is zero - does this mean the faces are tangent????

If I loft the whole, trim in half and then mirror I know it's symetric, but
often I can ensure that it's tangent.

Thanks in advance for your coments.

Lee

 

I try to avoid splines. SW's spline mojo isn't quite where it should
be.

If there aren't too many defining points, I find it is better to use an
ellipse or parabola to get what I want. Ellipses are especially handy
for defining fillets and scallops.

Overall, surfaces generated with more analytic geometry tend to hold up
better for many things, including holding symmetry. There is less
approximation and more true calculation of geometry.

The nature of lofts is that they are approximated surfaces. The more
precision you can build into their structure by using analytic curves
instead of splines, the better.

Also, try using sweeps with guide curves instead of lofts. Use a
pierce constraint to pin your section to the guide curves.

 

Lee, It is just not easy....

What I normally do is to define a guide curve on the mirror plane. Then
the other profiles defining the shape pierce this. Getting the
curvature comb looking "correct" for symmetry is critical and fussy. I
select the control arrow on the end point that intersects the mirror
plane, and constrain that to horizontal (or vertical) which helps
control the spline. It is harder to describe how the curvature comb
should look, but in principle you should image what the mirrored comb
would look like and ensure that that intersection has a continuous
character. And no, this is not ideal - there should be the ability to
define implied curvature as in Alias.

The next critical step is to ensure that when you define the loft, that
you select the guide curve on the mirror plane and define the condition
to be "normal to curve". If you do not do that, you have little chance
of having a decent mirror line.

I have also been asking for implied curvature / symmetry planes as that
would simplify the whole process for these types of products. Certainly
send it in as a feature request!

Daniel

 

There are several things you can do to help your symmetrical design. I
haven't found a single formula which tells you exactly what to do in all
situations.

1) the first thing you shouldn't do is avoid splines. Lines and arcs
give you K-cars. Splines give you Shelbys. Granted splines make it
easier to make bad geometry, so you just have to know what you're doing.
Unfortunately there is no replacement for competence.

2) Sometimes I model all the faces which do not cross the symmetry plane
on one side of the part, mirror those faces, and then create the faces
that go across the symmetry plane (using loft or fill). This is
probably the best way to ensure good geometry across the symmetry plane.

3) Remember that lofts and sweeps can go in two different directions.
If you can't get where you want to go in the first try, turn the problem
90 degrees and see if you can use GCs as Profiles and vice versa.

4) Tangency often is not enough. When you need curvature continuity
across the symmetry plane, then you either have to loft across the plane
or put a loft profile on the plane and use the curvature end condition.
Still, the curvature end condition requires you have a face on the other
side of the symm plane with the correct curvature, so you're kind of
stuck. This is why I usually use 2) above. The Fill surface, Sweep and
Loft GC will not allow you to specify a c2 boundary.

5) It's usually a problem that you can't mirror curves, so what I do is
use the curve to create some type of surface which can be mirrored, and
then use the edge of the mirrored surface as if it were a mirrored
curve.

matt

 

Didn't mean to say avoid splines entirely. Just that often there is a
better choice when an arc or line won't do. Especially true for
blends.

 

yeah, most of that (and the other suggestions) I've been trying.

Wouldn't mirror>curve be useful???? I end up doubling on loads of stuff -
reacreating lines etc. I like your suggestion about creating a surface and
mirroring that - but it's still lots of stuff for a simple thing!

Also, when you specify symetrical condition for a spline point it doesn't
maintain the symetry for the handles, and that would be useful too! (or at
least dont think it does, but I'll check!)

Thanks everyone!

 

Someone brought up curvature continuity across symmetry plane. One
thing to remember is that is a surface edge is coincident to the
symmetry plane and the surface is normal (perpendicular) to the
symmetry plane along that edge, curvature continuity is implicit when
that surface is mirrored.

I know it's not elegant to have mirrored surfaces with a line down the
symmetry plane, but sometimes it's the only way.

 

Yes, Mirror>Curve would be very useful, as would move, copy and rotate
curve, but since it doesn't exist, it's not too helpful for you at present.

Another way I mirror something like a projected curve is to make a derived
sketch of one of the sketches, and flip that using Modify Sketch, then
create another projected curve.

A mirrored 2D spline in SW05 should mirror the end point handles tangency
direction and weight. If you're using SW04 then it won't.

Splines in SW05 have improved a lot, but there is certainly still some room
for improvement.

matt