Simple shape for sweep2 as advised but why both options results bad?


I am bracing myself for flak and this being my ineptitude but until I get time to get at the lessons again, and that cant happen until I finish this job, I am to occasionally come across such simple things that dont go as hoped.

zebra shows A1 B1 CD missing as this is the area going wrong at the corners and across front.

I have this shape, it follows the same middle panel shape as Stratosfear had done for this object. making sure all sweeps are 4 edged. my outside edges are all 4edged sweeps to get best tight radius. very pleased. Then what appeared to be the simple last task has bitten me in the bum.

First sweep uses AB as rails and CDEF as profiles, the one edge is odd and curled.

method 2 split A and B into three and use as profiles.

result front panel again squiffy.

Its those radiused corners as per Stratosfear.

where should they best be, as rails or profiles, no doubt in this I have missed out on the right combination !

If I network srf this it goes well but its not allowed as it will mess up the filleting to be done after I slice across this, so I am led to believe.

Curvature Graphs are ok, please confirm.

Control point count same for the CDEF curves.

rest of shape has come together acceptably well for what I need it for.
Sweep2 whats best.3dm (154.3 KB)

I think I’s make the rounded corners meet at a hard angle, loft the section curves, then trim the rounded corners.

Sweep2 whats best_Loft.3dm (209.0 KB)

If the edges at the front and back do not hit come close enough to the curves you have there, use MatchSrf, for Position, to true it up to the curve.


I shall repost my post as I see yours just beat me to it :smile:Steve


Loft will create the simplest surface that goes through your curves. No need to add more constraints with rails, in this case it would probably add complexity and imperfections if your curves are not perfect. Have confidence in the math shaping the surfaces!
It’s good practice to use the simplest tool possible to get the job done and loft is the simplest while NetworkSrf is the most complex. BlendSrf, Sweep1, Sweep2 are in between. NetworkSrf has its uses but a lot less often than the others.
Trimming is not something to avoid. It’s part of of our toolset and the best way (sometimes the only way) to build some shapes. You have to understand what are the uses, the limitations, best practices.
So, answer 1; because the math at work to loft a surface between your curves will produce a smooth edge between the curves endpoints. Answer 2; try it, but in this case I think you won’t need it. Answer 3: it’s easy and fast to try it yourself. You can find a video of someone using Loft on a bunch of curves but you won’t know why your results are different.

I have extend curve the corners tweaked to get intersection and trimmed lines, ready to Loft, now as I have rails and profiles both of which arc, and Loft only uses profiles in one direction,

Q1)…how will it know the outer edges have to follow the two curves there ?

Have confidence in the math shaping the surfaces!
It got pretty close a sliver of light. but sweep would have followed as would networkSrf ?
filling that gap as you will see then has problems.

Q2)…is this where MatchSrf makes up for that ?

I find it comforting that surfacing follows rails.

Here is what happened, the corners wouldnt matchsrf, and by pretrimming them with the arcs I was unable to use MatchSrf. so pretrimming is out.

My 3dm file of the entire unit and the loft atached.

I have gone over it yet again and checked it with CurvatureGraph, tweaked anything in need and resurfaced all.
How am I doing ? any errors I didnt spot ? I dont have a tutor to check and cane me or praise me.

Then lofted it as per advice but found the corners gapped despite trying various settings.
Unable to trim with curve as then matchSrf doesnt operate, it needs untrimmed edges.

To see what lofting is about as I hardly ever use it because of so and to brush up on my Lofting knowledge to see what I should be considering more and get out of bad habits I google Rhino tutorials and get:-

I then type Loft in the search window as I want to go straight to a Loft tutorial, lasting a few mins if that, it finds nothing.

Q3) …Is there not rhino commands shown as a video per command, outside of the command help, (which doesnt always have a tutorial vid, and its a bit quick at times)
I dont want to have to plough through a water bottle tutorial hoping in there is a loft command. Not when the clock is ticking above my head on trying to finish something. Another time I will watch the entire tutorial.

I am not sure how Stratosfear did that panel but the curves and shape are the same.

I am trying to avoid trim as I have again established one shouldnt go trimming surfaces as they come back to bite one in the bum. Its all a question of knowing when its ok and when not. I hope thats explained in the lessons I will be starting out on.

Beginners would look at it and think, why not networkSrf or Sweep2, knowing what is best is the key knowledge.

Sweep2 whats best refinedCornerAndOneEdgeCurve.3dm (299.9 KB)


First, your curves are not clean enough:

You need PERFECT curves to get a good surface.

Hi Marc,
The profiles across from leading edge to trailing edge were without such. I then used extract isocurve on the initial created surface (I used networkSrf as Sweep2 gave a canvas skinned effect, even doing it strip by strip) extract isocurve as per Stratosfear method to get those rails inboard from the edges by 30mm or so and used those to split those aerofoil profiles giving the result, but then it has gone and messed up my lovely CurvatureGraph curves.

is that normal ?

attached are the originals before slicing. They didnt have those imperfections, they follow tabled data. data that was used to actually fly with. I tweaked the curvature graph for them as much as I dare.

are those CG imperfections responsible for the gap at corner ?

Addendum !..I also can explain the profile nearest tip of tailplane and CG at the ends, they were not like that, the ends are the extend curves I had to do following Pascals advice and of course as things go they didnt intersect, now I am not going to mess up the ACTUAL curve which was really good, so I turned on edit points and move the last point for the fore/aft curve a tad, to intersect, so as such not moving the curve that mattered. Of course this was now to give me a surface which wouldnt be perfect , but if I had moved more of the curve, then the surface and curvature wouldnt have corresponded to those other two in that area and not have been following the actual aerofoil dimensions anymore.

Comments on the original CGs and why after split command do such things occur ?
profiles before isocurve split.3dm (31.6 KB)


Yes, it’s normal. These curves are not appropriate to create a surface in one go.
The surface obtained from these probably wasn’t smooth and the curves extracted from it can’t be better than the surface.
“data that was used to actually fly with” is pretty much irrelevant. At the time the skin was hand-formed over the sections, the quality of the surfacing in a modern CAD has very little to do with your tabled data. And, working with real CAD from real flying airplanes, designed in this century, I can say that the shaded view smoothness requirements we are discussing here are more stringent than what we see in the “real life”.

Hi Marc,
certainly it seems like one is operating to something far more precise than required for the end result, in my case 1mm or even 2mm is ok for the current job, but one cannot relax standards as rhino wont accept it, tools fail, gaps result etc.

I find myself spending FAR more time just satisfying rhino than satisfying the tolerance of the job and as time is money its nuking me.

the amount of time into just a tailplane tip is 300% more than expected creating, CG checking , redrawing, then cutting and redrawing yet again thanks to the cut. Then tweaking to get intersections, then gaps appear.

i need to resolve that gap, but how ?

reach for networksrf ! or sweep2 and trim with curve ?


No, not at all.
If you start by simplifying your curves as much as possible (and without the arcs at the end), having a nice CurvatureGraph, having all the curves with similar point spacing by using rebuild to master curve, and then create a loft surface that is as simple as possible but stays close to you original curves (such as the one joined).
smooth surface.3dm (214.3 KB)
You can then continue you construction, based on strong grounds. It’s not a question of being less precise but of ensuring the input is as good as the output you expect.
On a surface such as this one you won’t have problems with trims or surface match, etc

when you say without the arcs at the end , you mean the large bulbous ends of curvature graph ?

I cant dispense with the bulbous ends of aerofoil sections :slight_smile:

That will mean redrawing (rebuild) the aerofoil sections, and rebuild does deviate from original curve shape, not just resite control points, it loses the curve accuracy, but I am also instructed to surface wings etc with them along lines that are drawn between their x/y plotted half cosine etc points. I am in a position to supply thw drawn sections to those that require them as per originals. ( I shall have to have another chat to my recipient, he might need a stiff drink !)

your surface is greenybrown by the way.

I see your surface goes beyond the central panel and down to leading edge, though it doesnt match it.

I also see that it overides into the 4 sided corner portion sweep that was Stratosfears approach, avoiding sweeps to points etc. I am now bemused as to if I do it this way, how do I get to skin the remainder as it looks like I am back to pointed sections. I spent some time, more time than I cold afford in fact, figuring out the 4 edge method, isocurves etc, and them messing up my profiles as you spotted. I need to keep my corner sweeps and sides etc. The leading edge misfit ?

I am with two approaches now and more than a little confused. I still havent confidence in loft having seen it not follow the edge rails. and matchSrf not fixing it without causing a massive gap. Somehow with time now almost having run out, I am not sure of my next move. I dont have time to rethink to this degree on this task.

I can try for the simplifying curves approach on the next task. The thing is, I am making an item from one plan and much later, a year maybe, I make another and the two must fir, working to the curves (not tracing them) on the dimensioned plans they were designed to fit 70 yrs ago, but if I rebuild to make the curves smooth, things will intersect and not abutt. I am unsure of that. If I supply my work to someone else who is also aviation precise, he might not be happy if things dont fit. its all a bit unsettling, pleasing rhino and deviating from original shapes.

I have tidied up the 4 profiles, only two wanted any attention, the one with one glitch I deleted an edit point and all is well, a 1mm square filled my screen and the before and after curve shows no difference aprat from one less edit point.

The one as explained with the extended curves for the Pascal corners, I nudged using 0.001 a few clicks and the down spike is now gone. reckon i gave it 0.004 nudge. as my final result is ok for +/- 1mm this sort of thing has me concerned given the time involved.


P.S see addendum in post following observation of poor CG’s for profiles, I realised why one profile was no longer perfect.

P.P.S I now see why things may be squiffy, I had 13 control points to my curves, by doing the Pascal corners they added in a few more, so now I have to try and keep this curve the exact same shape, else my top surface wont match that abutting it, rebuild I guess as longas its deviation from my original curve is within tolerance, it says 0.003 for 13 and deg3, so maybe I am ok.

Sorry, I didn’t read all. Your text is always long… :smile:

I just opened the file, and noticed a few things.

  1. The curve CPs are too many.
  2. Why do you create three surfaces to make your product?
    3.When you model something that has turning corner, don’t try to make it as a single surface. Think as a 4-edge surface first and trim that corner using blendcrv from adjacent edge curves.

Can you make your question very simple?


  1. I have mentioned the edit to the front curve , I saw it was 19, now used rebuild to make it 13 as per the other 3.

  2. I was following the method Stratosfear told me to do, he said 4 edged surfaces not 3 are best, see the metal looking iten part way down in :-
    VariableFilletSrf only partially selects my surfaces

  3. as per pascal, he said make it ‘sharp corner’ then trim afterwards. I thought seeing Stratosfears that the advantage was in the rounded nature of the corner, as I could see a definite advantage in the two corner panels where the profile swept around the approx 45mm dia curve, I used that same curve for the middle panel, ooops.

I created a 45mm dia circle then projected it to surface to create the curved rails rather than use BlendCrv.

However trimming off the corners to fit with those two corner panels afterwards disallows use of MatchSrf, so loft cannot be used. I have just tried networkSrf and got the best looking result, despite it not being an advisable command.

i wish this job was simple, it looked it, but following advice its gone a bit pear shaped !

My simple question is…if I use loft, how do I avoid that large gap when using matchSrf ?


Profile curves only need Degree=3 and 4 control points. No more or less in your case. Lofting or other surface-making tools won’t help. It’s not about what command to use, it’s about the topology of the shape. The big idea is making big surfaces first, then worry about details, the small parts.
Matchsrf, or anyother command, including VSR modeling tool, won’t help without understanding of topology. I will need your up-to-date files and make a demo of it.


Hi KevJin,
attached the basic starting kit ! all you need is glue and paint :smile:
Its the tailplane tip and a couple of profiles further inboard as indicators of how the top curvature continues on beyond the area the task is for. You will need to ensure the skin curvature is correct for its route to those hence their inclusion.

dont skin all of it, only that of C onwards to tip.

you will need some intro to it so here is that.

The item is same above and below centre so a simple task of mirroring the end result. The edge curves I have as sami cicles so once swept, a trim could be made. or halve them beforehand.

two profile sets included, thought I would include also those after I CurvatureGraph checked them.

the points are the only way to see the original dims tips when profiles transferred elsewhere (though knots can be used I have now established)

You may wish to start over with original profiles, you will see the Y dimensions, no use in including the X dims, they are from 70yr old tabled data and at the time dictated the curves. Redrawing what are flyable profiles without killing anyone will be interesting for me. (this job wont get to be built though for flight, but the principle and data of such is there !)

It will be useful and interesting to see how the integrity of the profiles and shape is kept.with 4 control points.
Not aware of such for aerofoil sections so eager to see.

Note the layer with radiused edges, quite small items ! radius gets bigger at tailplane tip.

Also there are the trailing edge radiused tips I created for the CG corrected profiles, as you see original profiles come blunt ended, these stuck out a bit beyond the trailing edge frame so I elected to alter the radii and keep things at trailing edge.

Hope all is explanatory. get back to me if not.

Stratosfears approach can be found at

The actual item needed further work as there is a radiused edge from aft edge round to just fwd of tailplane tip, he had it sharp. I followed his lead on the 4 edged surfaces method.

I am instructed by a very learned friend to skin wings etc strip by strip from rib 1 to last rib, the strips are the profiles divided up by the dim tip points Sweep2 , with rails being lines between those points. Then join all the surfaces up.

very good of you to be brave :slight_smile:
TailplaneTip starting data.3dm (175.6 KB)



Beautiful…I have a hell of a long way to go looking at that.
Interesting to see the different approaches by different modellers. e.e. that of Stratosfear compared to this.

It will take me a while to follow that video and understand the steps made.

Are there any tools etc there not native to Rhino V5 ?

Can you post the 3dm result so I can compare it to the original shape and profiles, also see how it cuts etc.

I am currently using rebuild to get all curves with same cv. however its coming up with curve changes, some quite alarming. made bullet tipped those radiused arc ends. one moment it wants 0.8mm then rerty a bit later and same curve 0.001, see my other post where no one has yet explained this.

Is that video shot in ‘real time’ ? You have brain patterns ans speed of Spock !



I tried upload the file here, but it exceeded the max size. I deleted the file.

I used VSR because some advanced features can save my time. If I had used Rhino only, It would have taken more time to do this, such as laying out control points and checking how surfaces flow. But the big idea is same.

I was wondering if I had to follow the original curves because those curves were not logically laid out. You should think about degree of curves and the number of control points. You don’t have to follow degree+1=CPs, but in your model I used Degree=7 +1 CP. Maybe you can also use Degree=6 curves or Degree=8. Demo didn’t redrew all curves independently, that would take me too long. That’s why I used orient3d to match up other curves approximately, but if you want to be more precise, you can adjust control points on top of “orient3ded” curves.

If you try to have a good result as you wish, you shouldn’t use rebuilt command. Adjusting CPs is always the best.

You should refine curves to match your original curves. In the image, I missed the word “degree” after curves.

By the way, even if you had the file, it won’t help because I just noticed one of my oriented curves ruined the model and that time I thought that’s impossible to make the shape you want because the control points on the shortest blue-color profile curve in the image (First one from left) didn’t follow the red line I drew. So, you will have to adjust curves a bit after orient.