HMAS Sydney II

[maurice]

Err .. yes, right, well .. as you correctly point out, a hyperbolic surface may be generated by and built from straight lines but it's not developable.
This leaves the notion of paper sagging but I would suggest probably only when going round a pair of concave and a pair of convex edge curves. Interesting.

As long as projection preserve dimensions and angles that's OK ..

In fact I'm accusing Pepa of not doing that but of altering (only very slightly) angles and lengths in and because of the initial projection process.
Giving sort of up to (but usually less than) a 1mm error in say a 400mm hull side.

BTW it is interesting qustion if this can be done analytically in non trivial case. It is easy to analytically unfold a cylinder or a cone - what about some other surfaces (generalized cylinders in this case).

If by the "non trivial case" you mean a developable surface generated by a straight line sweeping along 2 unequal and non similar "rail" cuves, then indeed it can.
Divide each unequal curve into an equal number of segments to produce quadrilaterals and triangulate those. Laborious in the extreme and although I've done it in CAD in simple instances I currently lack the knowledge to write any sort of script for the purpose.

[Roman]

Divide each unequal curve into an equal number of segments to produce quadrilaterals and triangulate those.

Well, by "analytically" I meant something else. What you described above I wrote my scrip for, and this I would call a numerical method as we are approximating curves with number of flat surfaces and than unfolding those.
Here is what I'm interested in: I can unfold a regular cylinder with my triangulation routine but why for? Why not just calculate the length of the unfolded surface (2*pi*r) and I'm done. No approximations required (well, assuming I can draw a line of length "pi"). So I wander if the same can be done for those "non trivial cases" you correctly described above.
I'm afraid I'm hijacking the thread from Mark – how is Sydney going?

[mabrown]

Hi Roman & Maurice,

Don't be concerned about hijacking of the thread. I don't consider this hijacking at all. This is interesting to me because I have been considering unfolding options since the very beginning of this project.

Some sort of unroll surface script (which I think you have alluded to with your regular cylinder example Roman) is the ideal I think. However, for those irregular cases, I'm thinking it would need to be able to determine the developability of the surface, perhaps determine at what point the unroll has exceeded a certain limit of developability and triangulate at that point, starting the unroll again. This might give a mathematically unrolled surface within developable segments. Just a half baked poorly thought out rambling.

MoI does very, very good triangulation of surfaces for polygon export. The level of triangulation is controllable in real time using sliders so that you can see how much your curves are being affected. I have found as a rule of thumb that a mesh which is generated with nice regular triangulation (as in my picture earlier) will unfold in Skoogle using the free ruby script. I believe something similar could be achieved within MoI using already existing functionality but there is no documented scripting or plugin api at present and the mesher only exports to disk; it does not retain a copy of the mesh for later use. None of these things are coming anytime soon.

So, I am left with modeling in MoI and unfolding in Skoogle. The potential problem I have left myself is that my frame is a pretty fair representation of Sydney's hull curves but my simplified, triangulated, unfolded skin isn't. How the two are going to glue together is in the lap of the paper Gods at the moment. I suspect that frame number one will be an unmitigated failure :(

[maurice]

So I wander if the same can be done for those "non trivial cases" you correctly described above.

I'm fairly sure the answer is no.:( Well not in the sense of there being a pi -like solution.
The Universe is just not constructed that way.:D

The potential problem I have left myself is that my frame is a pretty fair representation of Sydney's hull curves but my simplified, triangulated, unfolded skin isn't.

Not too much help I can offer here as there are different ways of doing things involved. MoI, and other modelling progs, intend from the start to produce an attractive 3D graphic and don't need to think about "unfoldability".
AutoCAD, and other early generation CAD progs, started out in life as 2D replacements for technical drawing with paper and pencil, only later moving into 3D and becoming able to do 3D graphics. With them it is still possible to either work solely in 2D or to move into 3D whilst continuing to work with them as technical drawing progs. This makes it possible to make the 3D product as if it were a paper model in which the 3D parts will be developable for sure. This is what I do and I suspect it is Roman's technique also.
Have you thought to split your hull into segments say between frames and use Pepakura to develop the shapes. Alternatively have you thought about Rollation as described in the Card Faq. This partcularly for the area between the stem and first frame.

[mabrown]

Originally Posted by maurice

Have you thought to split your hull into segments say between frames and use Pepakura to develop the shapes. Alternatively have you thought about Rollation as described in the Card Faq. This partcularly for the area between the stem and first frame.

Hi Maurice,

Many thanks for the rollation suggestion. Interesting technique which I wouldn't have thought of myself.

When I said "first frame" I actually meant the entire frame of the ship, this being my first and so far only attempt at such a design (actually design period). I'm pretty OK with the design from the bow to about 3/4 of the way down the hull. Things go awry the closer it gets to the stern. At the bow I did quite a lot of simplification which I've pig-headedly failed to do at the stern.

I am working at this moment on a plan to save the currently completed test frame. It will involve a little hacking I will let you know how this goes one way or the other (even if the photos turn out to be frightening).