EdgeSrf: Order of Selection matters, any rules behind how the surfaces are made?

While trying to make a surfaces, I noticed in EdgeSrf the order of selection matters in creating the surface.
Is there rules behind how the surfaces are made? Such as which one gets priority for continuity,
or getting tangent surfaces?
Attached is an example, and for some the tangency doesn’t seem to be kept when mirrored.

If there’s other rules for other command where selection order matters.


EdgeSrf_Selection_Order.3dm (1.1 MB)

The order of edge selection only matters with 3 edges. With 3 edges the V isolines converge where the second and third edges selected meet as can be seen in your illustrations. If any edges need to be extended using EdgeSrf then select that edge first. Next priority would be for surface matching. Select the most important edge for surface matching first. Also experiment. You can always back up a couple of steps using Undo and try various alternatives.

I’d add that you get the best (simplest, cleanest surface) results from EdgeSrf if the opposite pairs of curves match in structure.

And avoid having neighboring curves tangent to one another.

-Pascal

Thank you for the tips both!
@davidcockey I think I got it. So EdgeSrf is sort of several matchsrf done at once.
So what ever surface edge selected first stretches to the second selected then that gets stretched to the third surface edge.
This explains why the tangency fails in cases where I don’t select the frame edge first in this example. Even though rhino tries to make all srf continuous as much as possible.

I’ll try out different ways and try to get used to how the output goes.
I guess similar principle applies to networksrf with three sides, or patch.

@pascal When you mention: [quote=“pascal, post:3, topic:29498”]
if the opposite pairs of curves match in structure.
[/quote] , do you mean looking at the curvaturegraph?

No. I mean that the curves should match in their ‘internal’ structure, so to speak - e.g. let’s say two single span (degree + 1 points) curves will always work well - if you mix curves, say two different trimmed edges (generally complex curves) , the result will be extra complex in that direction.

The red EdgeSrfs are between the untrimmed edge (simple, single span, degree 3) & the trimmed edge (complex) of the curved surface in the foreground, and to a rebuilt (degree5/6 points) copy of the edge curve in the background.

-Pascal

Got it!
@pascal If there’s a need to connect the surface with different degree, should I use degree command to change the degree of the simpler surface to match with the more complex one?

Hi Toshiaki - the actual degree does not matter - the result will take the higher of the two if they are not the same - the key is the spans and knot distribution. An edge srf between a line (Degree 1 two points, in NURBS terms) and a degree three/four point curve (i.e. both single span) will be a degree 3 and four points - but if I randomly insert some knots in each, the result is a combination of the complexities, so to speak (points show knot locations):

-Pascal

Got it. At least an understanding… will try and see if my understanding is correct.

Thank you for the support!
The support level in this forum is pretty incredible.
Learned alot reading other posts and how you guys tackle each situation. :slight_smile: