Advice sought on when rebuild curve doesnt match result

Rhino Rhino for Windows
1

Hi,
V5
Being advised to revisit Levels 1 and 2 I have decided to lookin on those to get over this hurdle but cant find advice on what to do when rebuild doesnt match original curve.

I dont see anything in fact about creating a set of profiles using rebuild command keeping the shape within tolerance of the original. These profiles are aerofoil profiles and I dont want to deviate much from how they should be.

I need to sweep2 them and as I have been advised to ensure cv count matches, I have made them points 9 degree 3. They were already degree 3, so far the deviation has been within my limit I am willing to accept, but having done most, this curve will not match.

It is also advised that control points should be in same position for all profiles to get a nice sweep. rebuild places such in suitable spacing each time…as its an auto process so having this one fail me yet needing to match the others, I am halted in my tracks.

What should I do ?

I also see that whilst the original curve with tip drawn StartTangent had the control points where the tangent was, the result of rebuild places just one control point on the axis of the curve, so adjusting the tangent curve is no longer possible should I wish to do so. !

file attached.Rebuild no option matches curve.3dm (23.5 KB)

Steve

2

No, don’t use rebuilt command. The only time you would use rebuilt is while your profile curves are good, but you don’t want to redraw them. Which city do you live in the U.S? I think you need a tutor for advanced modeling. I’m in Los Angeles.

3

Hi,
I am in the UK.
I have no tutor, a 3 day course is beyond my pocket and I can see myself spending it just on profiles !
I have to finish this job before even considering such as I havent got paid for it yet !

I was told to use rebuild command to get all curves matching, I have in the past just drawn curves then swept them. Now I follow that advice so as to have the same number of points, it has done the job, except for two times it gave bad deviations then for noapparent reason offered a better one, then I get this one !

What command should I run so as to turn a set of profiles , lets say they are point counts 45, 16, 31, 13 and 9 into a set suitable for a sweep ? keeping the shape but following the rule of make sure the CV count matches. ?

I just wish I could draw curves following original dimensions (standard x and distance paired data from 70 yr old plans) and just use them , not all this messing about, adjusting things by nudges of .001 when the final result can be 1mm out and acceptable. Days spent fiddling about to satisfy rhino it seems. meals at 2am etc.

Steve

4

If you try to make a surface on top of point counts like you listed, the result will be bad. Perhaps single surface doesn’t matter, what if a surface that has lots of UV points and try to match adjacent surface(s)? What is going to happen is after using those match and blend commands, surfaces will have “jump” points that no one can handle.

5

Steve, I most likely don’t understand your question, but I’ll give it a try anyway.

The file you attached has two curves in it. Are these the curves you are trying to rebuild? This seems to be symmetrical, in which case you can just mirror one side and the result will be two identical sides, so I don’t understand what you need to rebuild.

To answer one part of your question, you can use FitCrv to fit a curve to a specific tolerance, but it will have variable point spacing. It is a tool that I very rarely use because the results are not usually what I was hoping for - the curves end up with unpleasant dips in them that produce crappy surfaces. I usually draw over curves or use Rebuild, but rebuild is not usually useful to me if the curve has a lot of variation in terms of it’s shape. For instance, in the case of your curves, they have a gentle curve and then an increasing curve as it gets to the tip. When you rebuild, you’ll end up with too many points in the gradual area and not enough in the tight area.

If you have a set of profiles that have various point counts and you want to rebuild them all to the same point count, I’d find the one profile with the greatest curve variation (if that’s the word) and rebuild that one to your tolerances, then use the same point count for all of the others.

Hope some of this helps.

peter

6

Generally use Rebuild to get the structure you want, not necessarily the shape - it is perfectly OK to rebuild a curve to a low point count and noodle the control points to make it match the original.

-Pascal

7

Hi,

As its for a surface across aerofoil sections it needs in this case to be one surface, also I am trying to get it to align to another feature so I am in need of maybe adjusting it and seeing how it then looks.
I think I am right in saying if I sweep with record history on, any tweaks to profiles will see an update of the surface ‘on the fly’ alowing me to see what I truly need.

Two curves as in a symmetrical shape made of two curves if exploded …
Yes symmetrical so in fact I could more sensibly unjoin the halves and concentrate on one half. In fact for the solution that occurred to me whilst shaving this morning, (sometimes bathrooms are good for thinking !), as the curves are aerofoil section with more happening at one end than the other, dealing with one half in the normal manner would be better. I overlooked the fact that they dont benefit from an ‘auto’ distribution of control points.

as you say…

you also noted the same task.
What I need to do is either take a point spacing pattern for the longest one and create a template and scale it down to the others, or create my own if none have what is called a half cosine spacing.
Having done that, as you also mentioned, redraw the curve (InterpCrv I shall use) with clicks on this template where it intersects with the curve.

again as you say

I wonder what should be the number of points in my template if I have to make one, I guess the same as for one of my other aerofoils where I have done this for.

I made the mistake of following advice of ensure point counts match and just jumping for an easy option tool where one could enter a value and have this done ! Another lesson learnt, rebuild tool is not so good if at all on such aerofoil curves.

Rebuild tool needs an extra option, a half cosine and full cosine point distribution option combo list for curves in need of more points at one end.

As such, shapes such as aerofoil sections will instead require drawing by hand, and if there are a series of profiles to be made to a standard structure with the same number of control points and same degree, use of a point spacing template for drawing them is instead the tool along with InterpCrv.

Having looked in on levels 1 and 2 this sort of answer wasnt there, unless it is naturally forthcoming once not just rebuild but the full course is under ones belt.

I did have the idea of an aviation side room to the forum where such tasks could be found so saving time beforehand.

Steve

8

Hi again,
Creating a suitable matching set of profiles for a single sweep.
attached here is what all this is about. task, to create a surface across this set of curves.
Source data has a mix of aerofoil sections (theoretical outside edge and frame data), and frames.

blue curves are frames which ARE NOT the outer surface of the tailplane, there is a skin thickness to be added on !
My plan is to get them as decent curves then offset them by that thickness, approx 1mm.

Red curves are theoretical aerofoil section which should one would think be the final surface. Where a profile isn’t the length I need I refer to another set that travel the full distance across this elevator and beyond over the tailplane. Elevator has a different aerofoil section in its forward part.

Some mismatching can be seen as the chap that drew the frames up didnt always consult the aerofoil plans or the chap that drew those ! My task to get a best fit surface for this area swept with rails being base line and sloping line. sweep profiles dont need to touch a rail I read, ! I cant add in a rail at base between ends of profiles as I am doing the sweep to find that out !

I have now introduced the aerofoil spacing template (white) which has 18 lines for mouse click points when using InterpCrv but as one can see, the radiused tip of the curve to be rebuilt (id#3) requires more than one line.

This template will be ok for some of the profiles but not all.

If I were to use this on numbers 12, 11 9 and 8 with their more typical aerofoil ends it would be ok.

lets say I have done that, then I come to #3, it will need an actual radiused arc drawing, then I use InterpCrv to continue, joining the two together (join or match command ?..) so I end up with a mismatched set of curve points. 18 for some and 20 for others.

I can see it will be best to trim the longer ones of 6,3 and 10 to the base line before using the template.
Probably also best to make upright 11 and 12 by proj to cplane then resiting them at their bases.

Please advise on best approach given the leader notes/questions in attached file and the task of dealing with radiused ‘tips’.

Maybe the radiused tips curve dictates the spacing for the rest, i.e tackle the most complex first and use its mouse click template to then create a template for the others ?

( I bet this isnt in levels 1 and 2.)
BestApproachAerofoilCurves.3dm (62.1 KB)

Steve

9

No - if the goal is to make the offsets good for surfacing, Offset first, then MakeDecent the results - in general, offsetting freeform curves always results in a more complex and less fair curve than the input.

-Pascal

10

Another command which might be useful is RebuildCrvUniform

11

Hi,
Pascal…thanks, in that case I will offset those needing such (blue curves) first.

I am still left unknowing just what to do after that, having had the ‘correct’ idea of applying a half cosine style template to the curves, there is the situation when the start of the curve is radiused, InterpCrv doesnt draw radiused curves, and constructing one and adding it in will alter the point count and point spacing, and all curves are supposed to match structure wise.

All I can think of is to construct the most control point demanding one of all then use that as a template to decide on control point spacing for the rest.

Is that the correct way …anyone ??

a whole day spent at PC getting nowhere on this job, thought i had the correct approach as well :frowning: it would have been but for the radiused tip ones.

david, cheers, could be useful, its actually RebuildCrvnonUniform.

12

I just kind of skimmed through this thread, with not a lot of time to read through it, so this might be completely off but from what I read you might be interested in trying the Rebuild to master curve.

You need to type in the Rebuild command with a dash, like this: -Rebuild

You’ll get an added command line option SelectMasterCurve. This will allow you to rebuild your curves to match the ‘master’ curve’s ctrl pt count, etc.

Hope this helps! Good luck!

Ciao

13

Hi,
a hidden extra function, if only there were videos on this and all that rebuild has, but then others say beware.

Does this mean it will follow the point distribution of the master curves control points , e.g if ;points are 1, 2 4 8 15 30 60 90 100% along the aerofoil chord axis on the master, they will be that same spacing along a chosen curve after its been applied, even if that curve is half the length of the master curve ?
Any videos on this…or more info ?

My problem is I have curves which I need to get all to the same structure and CV count (control point count)…I get told to make sure CV count is same for all, I get to see control points but no mention of what cv means, I assume cv = control points. I googled Cv count and found nothing.

I dont have a master curve with my ideal structure or control points. Not in the actual task. see that attached file.
I could use another from a similar file as its also aerofoil and the same aircraft.

I just need to get them all with the same cv count and structure.

I am then further worried by those with radiused ends as that radius must be maintained after the curve with it is redrawn. Do I need to draw those first then use that for my template ? Now I wonder, do I use that as my master , is it ok if it is with radiused end and other target curves have a different shape ? They must all retain their shape, I am not after altering them to its shape, just getting them all to have their control points in the same distribution as is the correct thing to do before sweeping them.

So lets try this.
I take an aerofoil section which has the usual point distribution which appears to be sort of half cosine. I take another curve and redraw it with a typical messy array of points, I also take a decent one and give it a lot of points, the measles !

see attached file.

I type _Rebuild and hit enter, choose my measled curve, hit enter and a panel appears, the usual one, I see no option for selecting a master curve though :frowning:

How should this work on my existing file or on this test file a video jing or whatever…

attached my test file with comments etc.
use of _Rebuild master curve option.3dm (71.6 KB)
Steve

14

Why not try it and see? It is surely less effort than typing a bunch and hoping someone else does it for you, right? I would not expect any command in Rhino to pay any particular attention to the requirements of any specific discipline - Rebuild does not know about the spacing along the chord axis of your inputs, for example.

-Pascal

15

You seem to be overthinking this whole thing. You are using 70 year old plans that were done with probably very crude means and expect clean results. You really don’t want to be recreating drafting errors from your reference drawings. I mostly do marine design but have dabbled with aero but when I first started trying to create boat hulls from sections and waterlines, I soon learned that too many input curves will soon bog one down. Does your project have to actually fly or will it be used for a printed model? The end use will determine the modeling strategy.

16

Hi Pascal,

but I do !..I go on to try it, rather than just ponder, and it fails. I just wanted to know what it should be doing if I got it to work. No harm in that.

Jodyc111…I am only after doing here what I have been told and that is if one has a few profiles , make them all the same point count. whether they are 70yr old plans or not is immaterial here, I am after principle of good matching profiles as told to do, but at the moment without a way of achieving them.

attached is the current suggested method, I trace with my points spacing from another profile, the uppermost (my actual task) curve and it fails. I need to keep the radiused tip as the edge outboard of that item is radiused. Having carefully drawn this shape, should one be expected to lose a radiused edge and have all sorts of deformation occur in the sweep ? Someone would soon say my profile was sloppily drawn and I have lost the perfect radius.

I am just trying to establish best way of getting all profiles in the sweep to conform to the rules I have been told they should conform to. that of matching point count and structure.

Am I right in saying I had best create the template that creates a match for the radiused tip curve then use that on the lesser profiles, its bound to look after their shapes ?

Steve

17

Did the command fail to do anything, or did it fail to do what you expect? If the former, please post the inputs and I’ll take a look.

-Pascal

18

Hi,
I select my curve, type _Rebuild and hit enter and I see no option to choose a master curve.

here is a screen capture.

Try it for yourself on my target curve that I used, the master curve is indicated in the file.

It would be interesting to see if it retained the all important radiused tip, certainly as you can also see from my trace of it using the template from the master scaled 1d, that approach failed.

i just wish to know from someone if its the right approach to create a template with more lines in that area or just even add one necessary to get a radius drawn then use that template for the others. Seems to make sense to me.

Steve

19

You need a hyphen, not an underscore, so -Rebuild

Sam

20

Dash… not the same as underscore…

-Pascal