this only works for planar profiles. The moment you crown the profile top surfaces and you look not from top view you may notice a weird flow.
So that is why I said, straight or jagged . If you create a polyline and not a curve at the intersection points this problem will disappear. However if you assume staying planar and you are not crowning the surfaces in z direction you won’t have that problem and you can use a curve.
This however is a very special case and won’t count for nonplanar profiles. I may code this case up, later the day since this case is rather simple.
No.
No.
It’s a purely intuitive thing for me as well. At the kink the two sections have to be perfectly aligned, and they have to remain perfectly aligned as they both slide away from the kink along different curved paths. You are of course free to rotate the section around the path tangent in any continuous fashion you like, but that is only a single degree of freedom, while the paths can curve away from each other in two different directions (only in one direction if they are co-planar, which is I’m guessing why it works in that degenerate case). 2-1 = 1 problem left you have no means to resolve.
Those look like special cases, i.e. the same curvature or balanced curvature on both sides of the kink.
Actually they are not! (one side is curved and the other is straight) Which accounts for non symmetric paths.
Redefines kitsch or the art of pointless or form defies function or I’ve missed something?
Stay on the subject Peter… I’m not trying to built a church (which I’m sure the L.O.D despices 
)
The interesting thing here is to find the principles of meeting sweeps.
And are you sure the curved side is curved in a plane that is not co-planar with the line on the other side?
To be honest no, I haven’t been able to find an elevetion of the building (or an approximating photo) But even if it is not the case there are many examples in baroque architecture.
You are European no? (I think Austrian) You must have visited palaces with ornamental works relevant to the subject. You can see them in painting frames/ friezes/ ornaments.
here! (you never can find easilly what you’re looking for when you want it)
If you notice the Cornice where the half dome meets the wall You will notice there is a double curvature.
Nope.
Eindhoven?
Note: with reference to the Useless Rhino thread: there are no curves here - it’s a picture, only bits and bytes…
So let’s make sure we are talking about the same thing. There’s two curves A and B, either non-planar, or at least not co-planar. A and B share an end-point, but no tangents at this location. There’s a single, planar, cross-section curve C. From A\times C we create a sweep surface S_A. We are allowed to copy C as often as we like along A, and we may choose an orientation of C at every instance, provided the normal of the plane of C is parallel to the tangent of A at the section insertion point.
Same for B and C.
We then want to slice through S_A and S_B with a single smooth surface, in such a way that there are no gaps between these surfaces left.
I’ve made a simple example. non-planar-joint.3dm (85.0 KB)
A (the red curve) and B (the blue curve) are both planar since they only consist of three control-points each, but they are not co-planar. The section is a straightforward rectangle. The red and blue dashed lines are the tangents at the end-point, the green dashed geometry represents the plane, normal and tangent bisector at the kink.
I’ve picked an orientation and created S_A (the pink surface) using _Sweep1.
the orientation of the blue section at the end of B has to be identical to the orientation of the red section at the end of A. We do not have any freedom here I think. I’ve drawn a second blue section some way along B, along with the tangent of B. We need to find a rotation angle of this second section around the tangent so that the long edges both intersect the creases of S_A. I marked these intersection points with dots x_1 and x_2. It’s pretty obvious that no rotation angle would possibly result in both x_1 and x_2 being true intersections. So what degrees of freedom do we have left to make this work?
1 Like
Finally some serious talking!
I couldn’t open your file (R5) but I recreated it in GH and indeed there was no possible solution.
I even tried to ‘cheat’ by putting one of the profile’s corners on the curves (thus guarranteeing that at least one of the creases would meet) and still… the crease’s distances would reach zero but never simultaneously.
bisecting surfaces.gh (32.8 KB)
I think we can safely put it to bed
Though some brave soul could investigate what pairs of curves could meet the criteria so that it can happen. I’m sure symmetry is not the only way…
Either way, I’m not so brave (or skilled) but the quest for the PLANAR kinky sweep continues…
Thanks a lot for saving me from pointless searching!
Possibly the stone masons did some manual modeling, and did not rely entirely on simple sweeps.
2 Likes
Dutch, but I do live in Tirol.
Eine minuten bitte!
I was in a hurry to kill it:
There is one condition that if met, you could have a non planar intersection: the paths just have to share one of their two curvatures.
As seen in this example, (where by definition the curves have one curvature in common as they are produced by intersection with the same surface) it works fine!
(by fine, I mean that the creases reached down to a third decimal away from zero)
Could this just be the impressisions that are due to the brep construction?
It looks convincing no?
and for every rotation of the one sweep, there is always a rotation for the other to bring the distances down to these decimals.
kinky1.gh (36.6 KB)
Actually, a ‘trick’ that is often used in these cases is to just sweep the profile at the intersection along the second path. The result looks great to the eye, but if you were to cut an intersection at each sweep you would get different profiles.
…And what’s more, those optimal points coincide with the intersections of the concentric ‘tubes’ of the paths!
kinky2.gh (47.4 KB)
1 Like
as promised, here is a simple planar case, however i haven’t spend much time on it, so I was lazy with error prevention,stability and code abstraction. However never indented to spend more than 30 minutes on this, so see it in this context as well.
Btw, I never said its impossible to have a curved intersection, i just said its bad practice. Technically seen its no sweep anymore.
Furthermore this approach is not how I would do it manually in Icem Surf, since i cannot reproduce the tools needed for mirroring that approach.
You also should not assume, 200 years ago they have “sweeped” the profile. You don’t know how and why they made i like this. Handcrafter often cheat or need to do compromises, but this does not mean its the best way in doing thinks like this. second image shows the problem with curved intersections when looking from perspective, the edge will have slight turning, you wont have that with jagged/straight intersections
planar case.gh (14.6 KB)
1 Like