Hull Design Software Based on NURBS

Discussion in 'Software' started by ancient kayaker, Jan 12, 2014.

  1. Mike Graham
    Joined: Feb 2013
    Posts: 67
    Likes: 6, Points: 0, Legacy Rep: 54
    Location: Maryland

    Mike Graham Junior Member

    The technical definitions are addressed, but I wanted to add to this thread some of the reasons why NURBS are NURBS.

    The non-uniform nature of NURBS is useful in giving local refinement on a continuous surface. This arose from the needs of freeform modeling, such as a ship or aircraft hull or a car body. It's more useful for design-y stuff than engineer-y stuff, but it allows very compact representation of very curvy shapes.

    The "rational" (which should really perhaps be called 'weighted') nature of NUWBS arose because plain b-splines were unable to represent circles and ellipses exactly. Rather than approximation, weights at control points allow these common features to be exact.

    B-splines are, as you hint at, 1D. 2D surfaces are made mostly by a cartesian product of two 1D splines. This could work in 3D as well (and there are occasions where this is done), but this isn't very robust at representing shapes that don't look anything like cubes. Because of this, the main way they are used for 3D modeling is through boundary representations, where several 2D patches are stitched together to form a solid. This can be finnicky, but it's more powerful than its main competitor, Constructive Solid Geometry, where only primitives with nice, clean analytical representation can be used. The most popular CSG programs are really NURBS B-rep programs with CSG as one way to operate.

    I always remain hopeful for improvements in the future, but it's difficult for me to imagine a model more robust than some sort of b-rep that is still very useful for solid modeling.
     
  2. The Loftsman
    Joined: Jun 2010
    Posts: 91
    Likes: 4, Points: 0, Legacy Rep: 41
    Location: The Loft

    The Loftsman The Loftsman

    Copenhagen Ships Curves.

    Hi,
    Just to be a bit pedantic, I have never used "French curves" in running lines although I have used extensively Copenhagen Ships Curves, along with Lofting weights and battens. While fairing lines I can assure you it required many more Lofting weights than 4 although I see that you use this as an eulogy.
    Although interesting subject for sure and it is all about what works for you I guess.


     
  3. ancient kayaker
    Joined: Aug 2006
    Posts: 3,497
    Likes: 147, Points: 0, Legacy Rep: 2291
    Location: Alliston, Ontario, Canada

    ancient kayaker aka Terry Haines

    Sukisol - you are correct. The curves I want to create are mostly easy to generate mathematically created with a spreadsheet or any software that can accept a formula as an input. FreeShip does not allow that unfortunately.

    Mike - thanks for the extra detail. I'm not sure what you mean by 1D splines though, sounds like a contradiction in terms, perhaps because I tend to think in cartesian terms.

    Loftsman - I used the term French curves" but I have heard of Ships Curves several times over the years, never seen a set though. Sounds like a nice thing to have, along with ducks and battens. For design I build I like to ensure the planks bend to design shape without too many station molds, that also makes for an easy build.

    Regarding the number of ducks required to define a given shape, there are perhaps more limitations on what can easily be done using a simple spreadsheet than with the standard drawing tools of a NA. I certainly didn't mean to suggest that 4 ducks will do any curve . . .
     
  4. Mike Graham
    Joined: Feb 2013
    Posts: 67
    Likes: 6, Points: 0, Legacy Rep: 54
    Location: Maryland

    Mike Graham Junior Member

    Sorry, I was not so clear. By 1D splines, I mean a line in 3D space, instead of a surface (2D) in 3D space. It's 1D because there's one parametric variable, even though it parameterizes 3D space. (You can also take the cartesian product of three splines to get a true spline surface; this is used in some places, but it's not used for standard 3D solid NURBS geometry, which is just comprised of all the boundary surfaces of the solid stitched together.)

    [​IMG]
     

    Attached Files:

  5. DCockey
    Joined: Oct 2009
    Posts: 5,229
    Likes: 634, Points: 113, Legacy Rep: 1485
    Location: Midcoast Maine

    DCockey Senior Member

  6. Mike Graham
    Joined: Feb 2013
    Posts: 67
    Likes: 6, Points: 0, Legacy Rep: 54
    Location: Maryland

    Mike Graham Junior Member

    Neither have I, but it's my understanding that it's all conic sections (circles, ellipses, parabolas, and hyperbolas). (This is in contrast to French curves, which are clothoidal in shape.) It would be interesting if someone out there was bored enough to do some curve fits on a really nice set, but it would be more than a little boring work.
     
  7. The Loftsman
    Joined: Jun 2010
    Posts: 91
    Likes: 4, Points: 0, Legacy Rep: 41
    Location: The Loft

    The Loftsman The Loftsman

    Curves and such

    Hi Kayaker,
    Here what you are saying, as for curves here is a Blog site which has curves to the extreme along with all the math and formula you could ever look for, good luck with the builds,

    http://thegeometryofbending.blogspot.ca/


    Cheers


     
  8. ancient kayaker
    Joined: Aug 2006
    Posts: 3,497
    Likes: 147, Points: 0, Legacy Rep: 2291
    Location: Alliston, Ontario, Canada

    ancient kayaker aka Terry Haines

    Thanks for the link to an interesting site, Loftsman. Unfortunately Mr. Nettelbladt makes the same mistake I did, identifying case 2 as a clothoid, although he was careful to write "probably". In a clothoid the curvature (reciprocal of radius) varies as the distance along the curve (from the finger end) but in the case illustrated it varies as the distance of the fingertip from the anchoring hand, which is the curve I wish to identify. So far I merely have a piecewise approximation but I cannot indentify the geometric term for the curve - if indeed it has one.

    [​IMG]

    - p.s., your forum name intrigues me - are you a boat model maker by any chance?
     
    Last edited: Jan 31, 2014
  9. The Loftsman
    Joined: Jun 2010
    Posts: 91
    Likes: 4, Points: 0, Legacy Rep: 41
    Location: The Loft

    The Loftsman The Loftsman

    Ship Curve

    Hey Kayaker,
    My trade in the construction and designing of the surface shape of ships, primarily in the days before computers, (which makes me a bit of a dinosaur and traditionalist always looking for perfection that just is not there in the shipbuilding industry anymore) doing many new build ships is a Loftsman have a look at my website for some history on lofting its use and also how conics were used for the first time to design the P51 Mustang which managed to destroy more enemy aircraft than any other in the Second World War.
    http://www.leithshipyards.com/mould-loft/lofting.html?start=5

    In fact anything that moves and has shape requires to be Lofted just that most people see it as being Design nowadays.
    As for curves we never used any formula apart from some when doing such as cambers etc, by running and fairing lines you arrived at just the "Fair" line or shape that was required more of an art than science really, a line that was pleasing to the eye, unlike todays mathematically derived curves which in my humble opinion tend to make everything look the same, in fact I can look at most lines of Automobiles and tell you which software was used in the design process.
    Anyway there was a very complicated curve in days gone by which was known as an O.G. or as our American friends pronounced it Ogee, this curve and sometimes there were two of them, from the main deck to the foc'sle deck at the side and on occasion from the foc'sle deck to the fore peak deck again this was a means of joining the sides and making the aesthetics of the ships look just right, this curve was no formula but was derived by throwing the points through the required three views with sometime a true view drawn through the sections cut to arrive at the desired curve, don't know if this is any help but hope you find it of interest, as for model boats never had the time nor the patience, but do show some very nice ones on my website as I appreciate the craft required to do such.

    Cheers and remember that "Knowledge not passed on is Lost"



     
    Last edited: Jan 31, 2014
  10. ancient kayaker
    Joined: Aug 2006
    Posts: 3,497
    Likes: 147, Points: 0, Legacy Rep: 2291
    Location: Alliston, Ontario, Canada

    ancient kayaker aka Terry Haines

    To paraphrase your words slightly, I can look at most any Automobile and not even tell you which model or make it is . . .

    I very rarely draw anything - small muscle control in my hands ain't what it was. However when I put a boat into software it is effectively already designed, often using a spreadsheet to express the concept I am aiming for. The software just does the pretty pictures so I can view it at different angles and maybe confirm the plank developments are as intended.

    When I first discovered hull design software I tried designing boats with the software and it just doesn't work for me unless they are mindnumbingly simple.

    Usually I have a build concept I want to try, and often work backwards from that to get a design that will allow me to put the concept into effect. Lately I've been developing methods for transverse planking, for example. Currently I am finishing a semi-stripped canoe for which I adapted the Adirondack Guideboat building method but with radial ribs. Another was a modification of Rushton's Wee Lassie canoe, stretched, with a smooth hull rather than lapstrake, built similar to the S&G process but without stitches: I haven't started that one. For an earlier boat I wanted the utmost simplicity for the plank developments; worked out the offsets, imported them into FreeShip and was pleased to discover how nice the boat looked. It turned out as well in wood as it looked in software. So the software is, for me, a prototyping aid.
     
    Last edited: Feb 1, 2014
  11. Baltic Bandit

    Baltic Bandit Previous Member

    well conics were abandoned for nurbs because of the computation involved, but that was 30 years ago. weve got the cpu power now. wonder if that's what the oneshape guys are doing, cuz tthat would inherently integrate the parametric surface and the solid. these are the same guys that did Cosmic Blobs a decade ago
     
  12. Mike Graham
    Joined: Feb 2013
    Posts: 67
    Likes: 6, Points: 0, Legacy Rep: 54
    Location: Maryland

    Mike Graham Junior Member

    I'm a little confused what you're indicating. Conics were not abandoned for NURBS--quite the opposite, they're why we have NURBS. The R in NURBS stands for 'rational', meaning that each control point has a weight assigned to it. These weights are an improvement over non-weighted b-splines in that they allow very efficient representation of conics. (No one would use NURBS if you couldn't have circular arcs!)
     
  13. ancient kayaker
    Joined: Aug 2006
    Posts: 3,497
    Likes: 147, Points: 0, Legacy Rep: 2291
    Location: Alliston, Ontario, Canada

    ancient kayaker aka Terry Haines

    It's my understanding that with sufficient control points any curve can be represented using NURBS to within arbitrary error limits, however many mathematically defined curves such as conics cannot be precisely represented, there is always some error associated with the result. That doesn't render NURBS useless of course, not by any means, and I'm sure they haven't been abandoned!
     
  14. Baltic Bandit

    Baltic Bandit Previous Member

    NURBS represent an infinitely thin surface. so to do 3D solids behind them you essentially take the boundary representation and then fill in behind it with a second set of parametric equations. It makes toolpath calculations of non-intersecting surfaces easier to do. But what it means is that you build things in odd sequences that need to be trained and for optimal model efficiency are done slightly differently in different CAD programs

    A true Conics based 3D modeler could be a solid geometry modeler. IE you define the solid or the void itself and the Boundary Rep is simply an attribute of the core equation governing the solid. This makes intersect calcs, particularly surface to surface mathematics easier and more accurate to do (you don't need "fixups" as your Solids equation is your surface equation.
     

  15. Mike Graham
    Joined: Feb 2013
    Posts: 67
    Likes: 6, Points: 0, Legacy Rep: 54
    Location: Maryland

    Mike Graham Junior Member

    I think I'm not understanding what you're saying or that you are mistaken.

    NURBS curves can represent conic sections exactly. This is why NURBS were invented, and what makes them different from non-uniform b-splines. http://en.wikipedia.org/wiki/Non-uniform_rational_B-spline#Example:_a_circle gives the right control points and weights for representing a circle exactly with NURBS, if you want to see it.
     
Loading...
Forum posts represent the experience, opinion, and view of individual users. Boat Design Net does not necessarily endorse nor share the view of each individual post.
When making potentially dangerous or financial decisions, always employ and consult appropriate professionals. Your circumstances or experience may be different.