Better Lofts in 2008?

Lofts in SWks 2006 for me with 3-6 profiles on a path have a maddening
& often impossible to resolve twists created by unmanageable and
constantly increasing numbers of connectors.

Those similar lofts are barely manageable with ellipses as profiles,
but when it comes to trapezoidal or similar form with slight radii on
the sides, and worse with corner radii, it is a disaster. All of my
lofts tend to create shapes that might be described as being similar
to some of the faucet shapes shown in the Help files, though my guide
curves often have an inflection point.

Hence, I am wondering how users have found the control and ease of
construction of lofts in 2008.

I am due to visit my VAR tomorrow to try out 2008 with him, and any
hints would be appreciated.

TIA - Bo

 

Bo,

I don't have enough experience with 2008 to say, but Craig used quite a few
lofts to generate the outside surfaces on a new product we are just getting
started on and I didn't hear too much swearing from his direction. We moved
to 2008 from 2007 because we thought it would be better with surfacing, just
as we moved from 2006 to 2007 for the same reason, but the improvements from
2006 to 2007 were not so stunning that I can recall any great triumphs. We
seem to have less trouble with surfaces now than we did with 2006, but we've
got lots more experience, so who is to say how much of the improvement is in
the software?

Interestingly enough, Craig ended up using mostly lofts, with only one
boundary surface and a couple of fills. We had thought going in that it
would be mostly boundary surfaces, but it turned out that the lofts seemed
to give him better surfaces in most cases. (He was duplicating surfaces
imported from Alias.)

Jerry Steiger

 

It is encouraging that he used mostly lofts.

What bothers me in SWks 2006, is that what would appear to be a simple
loft with centerline, where profile sections all have the same number
of elements in each sketch and all of the same types of lines and
arcs, and are relatively remaining proportional from profile to
profile should not automatically create a perfect loft with no further
editing.

I can see wild changes in profiles, and where radii wind up being
eliminated from profile to profile causing hiccups in loft formation.

I have hope that SWks 2008 has improved considerably.

Bo

 

perhaps you should take along examples of the situations that don't work
well in 06 and ask him to model it in 08 tools

hearsay in beta tells us the loft we are accustomed to may be less desirable
quality wise than boundary into the future...

 

"boundary into the future" means what? Am I enjoying so much
California hedonism that I've missed out on the boundary invasion
(said only half in jest)?

I do have myself set up with both the SWks 2006 sample solids, and
then the profiles & paths for those solids ready to execute in SWks
2008, so I can see what happens in trying to "correct" a 2006 model &
then what happens when I try to construct a 2008 solid with the same
set of profiles and paths.

Thanks - Bo

 

'boundary' is apparently the new 'loft' in waiting.
i.e. (if I understand it correctly) it is loft v2, but it hasn't been
publicised that way while the code coexists...
you could talk to your VAR about beta 09 and beyond...

 

Went to my VAR, but the loft issues in SWks 2006 are still the same in
SWks 2008.

When you get too many connector lines that SWks generates in a loft,
they become unmanageable, you can't position them right or delete
them, and they keep multiplying like rabbits and they get the loft
twisted into weird shapes, and at times you move a connector and the
VARs Dell goes into 30+ seconds of thinking before it redraws.

I think I've found that you have to keep the number of connectors to 5
or fewer (4 vertices) (when you Rt Click and pick Show All
Connectors). At the least, I can get a simplified loft done in SWks
2006, and then fiddle to get corner radii. 6 vertices in a profile
section appears to be unmanageable. At this point that is strictly
annecdotal evidence about the connector lines, but I am suspecting it
will hold true. If you tried to loft an Octagon through a bunch of
sizes & section positions, my guess is that it probably won't work.
If you put corner radii on the Octagon, I know it wouldn't work.

Bo

 

Bo,
perhaps you could use an intermediate profile to define the shape a little
better as it progresses rather than rely on the connectors over a distance?

I agree that the connectors are something that needs more user control.
I have experienced some difficulty 'chasing' them around the profile...and
not being able to position them as I would like.

These frustrations are not confined to loft though but boundary as well.
things like:
-able to lock connectors position either end
-able to dimension their position or specify a % along the profile
-able to delete connectors
-able to have coincident ends of adjacent ones

in UI graphics dept
-have the connectors numbered inside the circle for reference
able to highlight and pick them from a RMB list if they are too close
together or confusing

if there is any thing else you (or others) can think of say so.

what I will do is cut and paste the suggestions into the beta forum
somewhere appropriate so that some programmers get to see it directly

Neil

 

I am actually using 6 sections to carefully shape the solid loft for
its use and slightly for esthetics. I also use a centerline through
the center of the 6 sections which are symmetric about the center.

Bo

 

Bo,

Could I look at the loft (I just the sections and gc or center curve
info) and see what I can do or manipulate?

.. (bored... sitting idle...)

 

On the way for Saturday fun.

Bo

 

Various people offered suggestions which can be said to be generic to
a lot of loft constructions, so here are the general suggestions that
Matt L, Paul S and others noted in replies to me:

1. Get rid of as many extra Profile sections as you can. In my case
I'll use all sections for constructing some extra details so they are
used for construction other than the loft, so I kept them.

2. When constructing a symmetrical loft profile use only one half &
then mirror the final loft. Mirrors are more reliable, than all the
extra horsepower needed to do twice the loft work.

3. Break an unrully loft into two or more lofts.

4. Avoid extra objects in profile sections, and particularly put on
corner radii after the loft is done.

5. Splines linking profile arcs will add control.

6. Connectors remain a bit flakey, but you learn how to manage them a
bit at the least. Delete Connectors rarely seems to work, but Undo
Connector Edit works for the last edit when things go bonkers.
Sometimes dragging a connector dot over another connector can get rid
of another connector as a "Poor Man's Delete". I find using "Reset
Connectors" is mandatory when gobs of new connectors appear.

7. Make the elements comprising each Profile the same as far as
possible for each Profile. Consistency pays dividends.

8. Turning off "merge tangent faces" may reduce overhead in
generating the loft making it easier to get a loft to work.

9. Surfacing may be the end result for complex shapes to gain
control.

10. Choose as few objects as possible when making Profile sections,
and keep them as consistent as possible from Profile to Profile.

Later - Bo

 

OK, just a few questions on connectors in lofts which the SolidWorks
programming crew needs to answer:

1. Given similar profiles, why shouldn't logically connected
connector/s be the NORM, when the loft first is created?

2. Why do connectors look better organized after you first do your
loft and then use "Reset Connectors"?

3. Why does "Delete Connector" not work?

4. Why does moving one Connector cause the creation or one or more
new connectors?

5. Why does a default connector "cross" the entire pair of profiles,
and not be dragable back to a normal complementary position?

6. Why should the least number of connectors not be used when you
choose to Reset Connectors?

This is not rocket science, but for users of things like Lofts with
connectors, it is obvious there are big holes in the code for the
logic of how connectors should work.

7. It has been suggested that you can abandon the basic solid
modeling & proceed to surface creation to do more complex lofts. The
question should be, why should a user need to move to a more complex
mode of construction than is really needed with the likes of a simple
logical understandable and reasonable loft?

I'll bet a good programmer on these issues would clear it up in fast
time.

Thanks - Bo

 

Bo,

I was going to give you my generic suggestions to help with lofts, but I've
got nothing to add to the list you already got. Makes sense, since I learned
them from the same sources!

Connectors are flaky, but it seems to depend a lot on the profiles.
Sometimes they work just the way you would expect, and other times they
become extremely hard to work with. I don't know what the secret is, but
following the suggestions you listed helps a lot. They also work best when
you can pick the profiles in the graphics window, since you set up the
connectors by where you pick on the profiles. When you have to pick from the
Feature Manager, life gets much tougher.

Jerry Steiger

 

No 7 is what concerns me a little about the proposed take over of the more
precise 'boundary' code...
I wonder if everyone really needs to be accurate all the time and therefore
work through a more complex tool requiring a more advanced understanding?
Perhaps there are a decent number of users out there who aren't pedantic
about surfacing and are quite happy to get a fairly good loft from A to B
( with the wayward connectors tamed..) without too much trouble.
True to form SW guys don't talk about this stuff with users even in beta but
I did wonder if we should be making loud noises now to retain a general
purpose tool in the kit.
My understanding is that the code is not that hot so SW are keen to move on
rather than improve it.
Perhaps they could gut the too hard stuff that doesn't work or integrate
well and fix the connectors to make loft more handy and functional

 

Neil, I agree entirely with you. Other software programmers agree,
too, judging from another arena.

Apple announced that the next version of Mac OS X would not have any
'new features', but that they would spend their time 'polishing the
Apple', by doing bug fixes, integration and refinement of the code
under the hood of Mac OS X, such that everything gets crisper, faster,
more intuitive, more capable and more stable. These are my words
after reading what came out after the Apple developer conference in
early June.

I truthfully think that SolidWorks is in the same development phase as
Mac OS X. SolidWorks has added lots of capability, grown rapidly, and
have a list of areas users have made their wishes for fixes and
changes known. Now seems like a great time to have a release cycle
where the little items get cleaned up as some of these "little items"
can cause a lot of wasted time for users.

In fact, SolidWorks was founded on the idea of lower cost, easy to use
3D CAD interface that any engineer or designer could use effectively
with minimal training. Cleaning up the little inconsistencies makes
big sense.

Bo

 

Extrusion, Rotation, Sweep & Loft are the four basic processes that
satisfy 99.9% of my design work.

KISS is as good an acronym as anything to explain why.

Getting good solid work done in my products is almost exclusively
basic prismatic shapes, because that is what works easily in a time
efficient manner. Surfacing is great when it is called for, but I'll
bet the majority of SolidWorks designers don't use it much (I
obviously can't prove that). Machinery & tooling design is certainly
in that category, since they rarely require surfacing work in any
significant way.

I will use surfacing in basic ways when the standard features fail on
some odd connection that adds real world function that the standard 4
methods can't implement. But I am not good at surfacing. You have to
be committed to spending a lot of time on it to get good. I would
probably contract with some other designer to do a design that
required a lot of surfacing to be done right, and would learn in the
process.

 

Your situation is what I imagine is the case for many users.

I think there is a still need for an easy general tool as well as something
more specific/precise for the high end surfers.

My understanding is that the underlying code of loft couldn't deliver the
results in particular ID people were asking for and they have redone it.
Great but I hope they are not too ready to dispose of the old soln without
considering the impact for un-demanding users.

The attraction of SW is that it is intuitive and easy to use.
I wonder if the new soln, alone, wouldn't be introducing a 'steepness' in
learning and use that users don't need and may not appreciate.

If they kept loft (with a bit of a clean up) and renamed it 'Simple loft'
and made boundary 'Precision loft' , say, the two could coexist to the
benefit of everyone and the intent would be clear.

Something like Deform belongs in the same easy basket with 'Simple loft' (I
guess its just a variant of the same code anyway..) but I would re-promote
that tool as being for live hinges etc. and whatever else you could think of
to mangle things with.

Let other innovation deal with the surface elasticity that ID people may
want to play with.

Coming from 2005 to 2008 has made me appreciate how much new capability has
been put into surfacing and splines.

SW are to be thanked for that however they need to keep an eye on the big
picture.

There are many types of work that people require CAD for.
Some of the stuff is quite specific like sheetmetal or ID curvy stuff but
also there are users who need only the basics done well ( although perhaps
in large assys..)

If they cant code things to deliver both precision and simplicity in one
tool they need do it in two tools focusing on the merits of each.

 

How are they dealing with constraints, parametrics & history?

Any such capability (as in "represent a model") is & was in the
underlying NURBS (or other) basis to begin with even if the vendor
did not add application tools for the users to create/manage them
(for various reasons, such as the issues with managing constraints,
parametrics, etc. or the core math needed to calculate the desired
results based on user inputs).