A WTF moment with surface match

Any notion of what happens here??

Hm… to me it looks like it f*ks-up.

Joke aside, it is difficult to tell since all i have to look at is the zebra stripes, and that doesn@t tell me much about the surfaces, but it looks like your corner surface has too much data too close to the edge, can you turn on the control points for the surfaces and post an image, or share the file?

here’s the model at that point. I’ve since run into other maddening issues trying to build a clean surface. Such as the SS below. The surfaces were made from curves and a two rail sweep yet for some reason the control points at one U direction are on the other side of the surface formed!

Model

ArgoFlex2.3dm.zip (1.3 MB)

The zebra stripe problem is due to the display mesh being too coarse. I use Rhino for Windows so I don’t know what the interface looks like in Rhino for Mac, but in Rhino for Windows I can fix it by clicking on Adust Mesh in the Zebra Options panel, and them moving the slider all the to the right to “More polygons”.

Apparent continuity problems between surfaces which don’t make sense are frequently due to too coarse a display mesh.

Is the ArgoFlex file supposed to be the same model as the screen shot??

Yes, I sent a copy of the model at that point in its creation.

Hola! something further with the match surface method. Thanks for the advice on the pull command. After getting the corner exactly as needed, I have to match the other swept surfaces to it. However when I do this the control points go out of alignment on the essentially straight sweeps.

I used the edge of the corner surface as the profile to sweep instead of the edge curve. Is this perhaps causing some mathematical issue? I do it that way thinking the sweep would more likely match the edge it was created with.

ArgoFlex3.3dm 5.zip (148.4 KB)

Hi James - in my opinion, this is exactly backwards - the straight surfaces are givens, right? They are exact and known. Make those. Then, monkey with the transition surfaces to to make them match up to the ‘hard’ surfaces.

-Pascal

I thought that way to and tied matching the surface of the corners to the straights but I get a pool seam, and as you can see another anomaly. I can figure at what point I’m inadvertently creating some bad math. The example below was made with all new curves, each matched to the straight rails and vertices. All are degree 5 curves.

ArgoFlex3.3dm 6.zip (418.9 KB)

When you MatchSrf, pick both edges clearly at one end, and the same end, of the edges or you’ll get a bowtie like that.

-Pascal

Still get problems. And I wonder why if one sweeps the edge of a surface along two rails the seam does not match as shown with the z stripes. It seems making a surface this way should automatically match up.

I’ll just point out a few things that I notice… make of it what you will

  • it looks like you’re not constrained on the corner shape. If that is the case I’d use simple G2 curves.
  • the existing fillet shapes are made from rational curves and have weighted control points - this adds an extra complication and should be avoided if you want to keep your surfaces simple and curvature continuous. My guess is you started from arcs (rational curves, weighted control points) and used ChangeDegree, possibly?
  • the corner surfaces seem to me to be far more complex than needed. Matching surfaces o curves moves up to 3 control points in from the edge/end. The more crowded, or closer to the edge these are (as in a dense surface) the more ‘local’ the change will be to the surface, leading to the wiggle in the zebra that you see.

ArgoFlex3_PG.3dm (444.0 KB)

-Pascal

Yes, I did this. I concluded it was useful from an earlier bit of instruction regarding the curvature rating being 5 making possible the use of edge surface method. I’ve since lost interest in that particular method in favor of the blend surface approach.

So, in principle one should reduce every input curve to the max necessary to describe it. In fact the edge curve I created from using blend and then pull to surface was only a 3 degree item and I actually upgraded it to match the others that had already been increased.

Is there an optimum order of surface matching steps - such as horizontal then vertical, or smaller then larger, etc?

BTW - are there any user examples of key combos setups regarding the osnaps?

As a rule, yes, that’s a nice way to work, where you can, with the caveat that if curves are to be combined to create a surface, they should match in structure. This can get a little tricky because it’s not just point count - a six point degree 5 curve does not match a six point degree 3 curve but it does match a two point degree 1 curve, or a four point degree 3 curve, assuming the curves are all uniform and non-rational. The point of matching structures is the number of spans, again assuming uniform knots. Number of control points - Degree = Spans.

(This is further complicated because only some surface tools care about structure, notably EdgeSrf, Loft, & Sweeps when combing curves in one direction or the other. NetworkSrf does not care, but also lacks the ability to make very simple surfaces in most cases.)

I don’t think there is a rule about matching order. But it is worth paying attention to the options for isocurve direction, closest points etc - these need fiddling per the situation.

-Pascal

Okay, that helps. I’ve still a question regarding matching up curves for optimum surface creation. In the ss below the hi lighted curve is as simple as it can be and still hold the shape it needs to make the surfaces that join at it’s seam. Given the point count and degree rating what would you rebuild or change to the other curves it will interact with to create surfaces?

I guess I would question that, first. Is this curve the same shape in plan as the base curve on the corner?

-Pascal

Yes. the curve in question was first made with a blend and was then pulled to a surface extruded up from the base curve so as to be absolutely vertical. I also had then matched that curve to the two straight horizontals used to created the blend in the first place and then pulled it again to the extruded surface.

Hi James - try this:

  • Copy the base curve from either end point to the corresponding end point above of the fillety surface edge.
  • Turn on points.
  • Select the three points at the other end of the curve and Move them from the end point to the other corresponding end point of the other fillety surface.
  • verify continuity with GCon, if in doubt.
  • View from above…

-Pascal

:relaxed:

Very nice. In fact, I was able to get the curve with just a total of three control points.

I’m getting an anomaly now that I can’t figure out. I made a vid. Every curve used for the surface creation has been matched to every straight line. But matching surfaces caused a disruption.

ArgoFlex3.3dm.zip (155.7 KB)

Update - worked through it again with new curves and matched curves and no matter what I keep getting this problem. I’ve looked into the direction of every curve and they seem to be correct. I’ve even tried reversing them but still it won’t behave. Attached is the surface after being matched to the two rail sweep. For some reason control points go out of whack.

There’s also an issue with dragging an edge from a surface and having it be inside out. Go to the 20 second point and watch please.

http://www.screencast.com/t/wSNZKBPKwhLt