Curvy Stuff 101 questions

I had a couple of questions on Curvy Stuff 101, slide 29. Was the
point of that slide to show that is easier to put limited loft
sections at the points on the spline versus many more sections at
regular intervals? Also, in the model, a sketch is named "predicts
total sections required for smooth loft". What predicted the 12
sections needed for the smooth loft. Thanks Pedro

Many thanks to Ed Eaton for making this information available. Great
job!

 

Pedro,
The point of the slide is to show that lofts are spline based, and that you
can use sketch splines to help you decide where to put your loft sections.

I wanted to show that if, like a lot of folks, you really insist on using
evenly spaced sections, you can use a sketch spline with evenly spaced nodes
(achieved in sketch mode by using a lot of centerlines with equal
relationships) to see just how many nodes/profiles are required to meet your
design intent. In the case I show, the loft would require 12 evenly spaced
sections. Its a lot easier to use a quick sketch spline to predict how many
sections are required than to try the loft with 8 or so sections, see that
it doesn't work, edit all 8 planes and profiles to add a ninth, see that
THAT doesn't work, then do it again and again until you get to the twelve
sections that will work.

I also use this technique when trying to follow two intent curves, like the
top and bottom curves in an elevation layout sketch. I will draw a sketch
spline on the top curve with as few spline nodes as possible, then draw
another with the same number of nodes and see if I can use that spline to
match the bottom curve (the line between corresponding spline nodes
indicating my loft profile planes). Of course, it never works the first
time - I have to add nodes and move them around until I get the right fit,
but at least I quickly end up knowing exactly where to put my loft profiles.
Also, I have to take into account the effects of angled profiles. Still, it
is tons faster to take a little time in a sketch to get things right than to
take a lot of time blindly editing, adding, subtracting, and repositioning
loft profiles.

But, of course, the real revelation of the slide is to point out that
relying exclusively on evenly spaced sections like we are taught is kind of
dumb (with the exception of simple models or industries where other factors
pop up, like bulkhead spacing or whatnot). Even the inventor of NURBS
surfaces told me that the concept of using evenly spaced loft sections is
inherently silly when applied to NURBS - the whole point is that these
surfaces are NON UNIFORM rational b-splines, not 'uniform' (URBS?) which is
what you would get with evenly spaced profiles.

By using non uniform profile spacing, we can take advantage of the math used
behind the scenes to meet the design intent in the sample with only five
profiles - thats a LOT less set up, and a lot less to modify for design
changes or tweak when fine tuning a model. Less work and better results? I
thought that was a revelation that was worth passing on.