Closed Loft

[RobertoMolinos]

Hi, I am trying to make a closed multisections surface in GSD, that is, with starting and final section being the same and thus having closed guides, and it seems not possible.
Any idea?
Thanks for any help.
Roberto

 

[Kapitan]

Without having an image to look at, I'd model whatever it is you're doing, in two halves and join it later. Split it like a big-end bearing.

Can you do an Tools=>Image=>Capture and post here?

If the start and end sections are identical, and co-incidental, it may be a simple surface of revolution solution...and not need guide curves.

 

[RobertoMolinos]

Hi again

Thanks for your reply Kapitan, I already tried that, what I need is a closed loft, not a revolution, because I have more sections than the one that is initial and final,and my guides are not necessarily circular
I dismissed doing it in two steps because I need G2 in all the surface, and I don´t know how to get that in GSD.
Follow the link to see a screenshot of something similar done in rhinoceros using loft with closed loft option selected

http://paramarch.homedns.org/data/closedloft.jpg

greetings!

 

[snowfire]

It is not possible. have to do it half and half.

Forever Young

 

[catiajim]

G2 (Tangency Continuity) is easy to do using the multi-section surface tool. Build some extrude surfaces at your start and mid-point profiles. They can be fairly short. Use these surfaces to control the tangency of each of your multi-section surfaces as you drive each half of the surface. Hide your extrudes, and join your two multi-section surfaces.

 

[Kapitan]

This closed surface appears to have no planes of symmetry, it wiil be as good as any section and guide curves, and maybe a spine, that are used. There is requirement for any connections to have G2 continuity - I've usually taken this to mean Curvature Continuty, the following is from the Help Documentation:

G0 If the endpoint of curve K1 meets the endpoint of curve K2 then we say: At this point both curves are connected with order of continuity G0.
If one edge of surface S1 meets an edge of the surface S2 then we say along this edge both surfaces are connected with the order of continuity G0.
If the G0-continuity is missed then we have a so called G0-error. This error is an absolute error, a distance, and it is measured in mm or inches.

G1 The curve K1 and the curve K2 are connected with the order of continuity G0 in the point P. If both curves in the point P run into the same direction, this means the angle between the tangents of both curves is 0, then we say the order of continuity is G1.
The surface S1 and the surface S2 are connected with the order of continuity G0 along the curve C. We take the normal of S1 in a point near the curve C and run with this normal over the border to S2. If the normal does not change its angle from one point of the border of S1 to the nearest point of S2 then we say the order of continuity is G1.
If the G1-continuity is missed then we have a so called G1-error. This error is an absolute error, an angle, and it is measured in deg of rad.

G2 The curve K1 and the curve K2 are connected with the order of continuity G1 in the point P. We look at the curvature vector of K1 in point P and the curvature vector of K2 in point P. If both vectors have the same direction and the same absolute value, then we say the order of continuity is G2.
The surface S1 and the surface S2 are connected with the order of continuity G1 along the curve K. If each curve on S1 which runs over the border to S2 can be continued with another curve on S2 and their order of continuity is G2 then we say both surfaces are connected with the order of continuity G2.
If the G2-continuity is missed then we have a so called G2-error. This error is a relative error and it is calculated with the following formula. K1 may have the radius R and K2 may have the radius r at the common point, with r<R, then yields:

This would be a simple task, if only Tangency Continuity (G1) was required, it's a bit more complicated than that to hold a G2 condition. I'm sure it can be done in GSD, maybe using Tangency then Curvature.

Perhaps the way to approach it would be to make the first half using three sections - each with tangency extrusions; then make the second half using start and end sections of the first half - and it's Tangency & Curvature, not the extrusions. Not forgetting the middle section - probably free of any tangency.

It may well be best to not use any guides, but a spine instead. Sometimes too many inputs don't give the best result, if guides are used then it could be important where they intersect the section curves.

An interesting problem Roberto, what are you starting with - just 4 sections? how were the guides constructed?

This shape reminds me of some mathematical surfaces that are defined by a parametric equation, an interesting exercise using VBA macros in Excel, and creating a Loft from an external file...

 

[RobertoMolinos]

Hi and thanks for your replies.
As kapitan says, it seems difficult to add G2 to both half lofts, I have tried the aux. extrusions and tangency_then_curvature and seems not to work:

Maybe I could build both surfaces in GSD in a proper way and then impose G2 or even G3 within Freestyle Workbench.
As an option, I moved to imagine&shape and got to something similar deforming a thorus, is not the perfect solution, but since I am checking CATIA/digitalproject capabilities in architecture, it fits me and gives me another point of view to model this dome:

Hope I can get to a good point with this and dont have to go back to rhino.

 

[sparck]

If you have 4 sections, you can try to do a multi-sections between 2 sections, then do it a multi-sections between the other 2 sections (like two oposite quarters), then make multi-sections in the gaps using 2 sections and surfaces (the first and second ones)like tangent supports.
I hope that you get a good result for what looking for.

Sparck

 

[Kapitan]

If the first half is made with Tangency only, as a G1 piece, and then the second half is made as a Tangency + Curvature from it, as a G2 piece, the first half then hidden and a third half made as a (G2) replacement first, it will take it's tangency and curvature from the second half.

Of course, this diminishing difference approach could go on and on, but if this part is to end up being manufactured, then the reality of machining a mould, or whatever, with a ball-nose cutter and then hand finishing means that one has to think - just how far do you go with all this....

Ideally, the model would be a single surface, but if that is not possible then the next best thing is two halves. Breaking the thing down to four parts is opening up opportunities for (maybe even) more errors.

 

[slcad]

How about making the lofted surface using all the sections, then make a blend (or two?) between the 1st and last section? Blend surface has the curvature constraint option.