Below are a series of shots showing the result of extending to lines from snapped to points that are supposed to be planar. I run into this problem often and I keep thinking it’s some misuse of the program by me, but it’s becoming very frustrating and I’m constantly wondering if I can rely on previously built items for accuracy down the line. As best I can at this point, I’m creating curves as precisely as I can yet I get nonsense like this.


If you are working with the NEAR snap on, turn it off. It is a major source of problems like this.

Additionally, it is important to realize that the Ortho command is a major hazard to rely on in Rhino since any snap will override expected orthogonal behavior. DO NOT RELY ON IT! (This is sort of like in the glorious old days when using an adjustable triangle or Mayline that is loose and the angle slips as you’re drawing a line.)

Instead, you will find the best success by drawing a line in an orthogonal view and immediately hitting TAB to constrain it to the desired vector, then you can set the endpoint to the desired location. Alternatively, draw a line anywhere and move it to the desire location. Turn Points On and move these to the desired snap location if adjustment is needed.

Is this good or desirable behavior? Hardly (…but good luck convincing died-in-the-wool Rhino users of this incredibly obvious problem for anyone who has used any other program other than AutoCAD). ~Dave

@JKayten I always seem to get this with your uploads… Can’t get the file(s). --Mitch

Mitch, Sorry about that - not uploading the model. I’ll make a point of doing so from now on. I don’t always imagine it’s necessary but it couldn’t hurt. The model has evolved and the state shown is no longer available for review.

I think one of the problems with snapping is locked objects. I make such snappable to use guides laid out. However, I also lock image files and those have edges that are not necessarily on the same lines as the layout. I wonder if there’s an option of making some locked objects absolutely locked while maintaining the usefulness of others.

I watch these training videos by Kyle Houchens that are very useful. One thing he repeats is the growing ability to actually design in 3D. It seems anything that facilitates or improves such an ability is the most positive direction to develop towards. I think there may be a few more tricks available in the windows version than currently in the Mac, but soon I hope they’ll have equal degrees of utility. I’m looking forward to the eventual plugin access the the T-splines world.

I’ve been whining about my pet peeve - construction lines for some months now. Still, one must adapt to what is. Thanks for your attention and help.

Okay, here’s another incidence. As per the image, I drew a polyline from an intersection and snapped to an intersection and yet the line overshoots the line it’s intended to intersect

Model attached (645.3 KB)


Hi James - intersections can be real, or ‘apparent’ - check settings for whether snaps should use apparent intersections (Modeling Aids).


Damn! that option was checked. that must be the default, I don’t recall choosing it.

Yeah, I have run into this before as well, there has to be some request by me back there somewhere for “unsnappable” reference objects for the same reason. The only way out of this for the moment is to use background bitmaps, but they are somewhat painful and unwieldy - plus they only show up in one viewport.


Here’s the latest. I’ve been ridiculously careful to make everything copacetic and yet I’m left with some annoyance. I want to split the upright curves with the highlighted horizontal but while it’s supposed to be intersecting the objects it’s apparently not since I can’t get a proper spitting.

Model attached (653.3 KB)

Yes, the curves miss each other by .004, which is above the file tolerance of .001. To construct this stuff accurately, one of the ways might be to create a plane surface at the appropriate horizontal level, intersect your vertical curves with the surface, run lines through the intersection points, then fillet the corners… At least the intersection points would be guaranteed that way. You can also just split the vertical curves with the surface if all you need are the split parts.


I actually need the horizontals to make a 2 rail sweep and I want the split lines to be resting on it so I get a clean joint for a merge surface to the top plane. I’m still struggling with this.

This the same model state sent you last

Hi James

There are always different methods to create the model, in this case, one set of the offending curves needs to be redone. The following method redraws the horizontal curves as the radii you’ve used are a fixed size.

Explode horizontal polycurve and delete the right hand line with it’s two fillets - which are 8 units radius each, so easy to redo.

Flip CPlane to Right,
ExtrudeCrv the left line to create construction surf,
Split the angled lines with that surf then delete the surf,
Draw 2 new horizontal lines to the newly split line end points,
Fillet the new horizontal lines with the existing at 8 units radius,
Using a combination of 0 Fillet, Join and split on new horizontal as req’d to create the wires to build the surfaces.

As a check, all the various construction curves should be able to 0-radius fillet with each other, then you know your input curves are tight before applying surfaces and you know you will be able to join the surfaces into a good polysurface.

So using the fillet tool at 0 is a means to check continuity?

Also, the corners you created look very smooth. Were those sweeps as well?

FWIW, I still think you can greatly simplify -

  • your vertical curves appear to want to be straight with a curvature-continuous transition to the top, and tangent at the top face. But they are not straight and not quite curvature continuous, and in fact reverse curvature in some places.

I would use straight lines/planes where the surfaces are meant to be planar and isolate the transition surfaces as separate surfaces.

ArgoFlex2_PG.3dm (1.4 MB)

Yeah, I’m not sure why I’m getting such issues as I try to make all as simple as possible. For instance the upper curved part of the verticals was made with a simple line drawn from two intersections in the guides and then converted to a 4 point curve. I then used match curve to make that curve tangent at each end and assumed that would be suitable. The curves you’re showing here I guess are the ones you’ve remedied in the posted model?

I checked all the curves and they’re clean, I think the model you used is an older version. I’m intrigued however by the curve you created. Is that a tool used or the result of some intersection of curves… ??

I placed points at the end of each curved section to straight on the corner vertices and then used curve though points but that didn’t get it.

If the radiused parts of the surfaces were blended with ‘Blendsrf’, that’s how the resulting surface edge would look.

A regular 0-Fillet check shows the curve end points are coincident before creating the surfaces.
The corners in the screenshot are 2-rail sweep from your wires, although another method for achieving a nice corner here would be ‘BlendSrf’.

You would lose the exact radius on the wire at the bottom edge of the blend, but maybe that’s not critical. Then use that wire and the bottom corner wire to create an ‘EdgeSrf’.

ArgoFlex2 BM.3dm (190.0 KB)

I made the blend curvature-continuous because I thought I might use the extra points to resolve the arcs and straights at the bottom more closely - in the end I added some more points to get things stuck, more or less, to that bottom curve. I always assume, where possible, that the file posted is what the user actually wants so I did not make that corner actually a G2 blend, I tried to get it close to the straight-arc-straight…


Okay, so I managed a couple surfaces to make a blend and this. I know it’s something to do with orientation, but what I don’t know is how I manage to get such things out of whack in the first place. Is there some setting that I’ve inadvertently mis set?