Could you clarify where your original data is coming from? Are you using tables of coordinates, or do you have some geometry to set up the curves, or are you tracing off a paper print, or a scan of one? I’ve done loads of similar work, in v3, v4 and v5, and never had pickled surfaces like that! Thanks in advance,
Tim
Steve -
Jim is right, you’re totally in over your head here, and you’re thinking that you’ve found bugs and problems in the software, when you really just have absolutely no clue what you’re doing. Seriously, you’re making an ass of yourself here. I do EXACTLY what you are trying to do for a living, and I’d rather saw my left arm off than help you at this point. JUST. STOP. Take a step back, and read through the Rhino Level I and II coursework. Or, take a class. I took Jim’s Level II class, it addresses pretty much everything you’re trying to understand here, but failing at.
So why is that sweep1 all messed up?
The curve on the left is made up of 4 segments. One is an elliptical arc segment, three are NURBs curves. Two of those NURBS curves are degree 3, one is degree 5. All of those curves have terrible point distribution.
The curve on the right is made up 1 degree 5 NURBS curve, again, terrible point distribution.
Your rail is made up of 6 segments, each degree 3 NURBS curves, with terrible point distribution.
Normally at this point someone would say something like “garbage in equals garbage out” but calling what you’ve done is just insulting to garbage. It’s more like “Ignorance in, bewilderment out.”
Seriously, go read the Level I and II coursework. No really. I’m not kidding. After you do that, then start thinking about accuracy.
-Sky
Here is a link to the Level II coursework
http://www.rhino3d.com/download/rhino/5.0/Rhino5Level2Training/
-Sky
Tim,…hi
I am working to ancient engineering drawings, plan and side view, with some dimensions.
3 8 5 6 9 can all look the same.
The numbers are sometimes legible. They oftwen look like someone sprinkled soot over the copier screen first. The tips of the arrows are not clear and sometimes one has to use assumption that they are to the inner curve of a profile section. Its not tabular data but plans of parts and also some GA drawings. Its also sometimes a question of scaling up a plan to what few dims are there and tracing off the shapes well aware they are not drawn to scale.
Airfoil data I had was tabular, another single plan was entirely woozy with 7 9 1 looking ths aem as well as 3 8 9 etc.
skyg…read again what I have posted, in it you will see the reason for the differing curves. as explained above. Different people drawing things up and not checking with each other. Ok back then but not RhinoV5 friendly.
So why is that sweep1 all messed up?
The curve on the left is made
up of 4 segments. One is an elliptical arc segment, three are NURBs
curves. Two of those NURBS curves are degree 3, one is degree 5. All
of those curves have terrible point distribution.
The curve on the right is made up 1 degree 5 NURBS curve, again, terrible point distribution.
Your rail is made up of 6 segments, each degree 3 NURBS curves, with terrible point distribution.
Normally at this point someone would say something like “garbage in
equals garbage out” but calling what you’ve done is just insulting to
garbage. It’s more like “Ignorance in, bewilderment out.”
I didnt realise drawing curves to dimensions on original plans would cause that odd set of curves.
Thing is they look ok to the naked eye and it was a naked eye 70 yrs ago that drew the curves then wrote down what dimensions were required to create them.
Rhino is something they never had back then. I am not in over my head. I am , or was, operating to an expectation to adhere to these dimensions.
Its the first time I have had a sweep go haywire and I have been using Rhino for quite a few yrs now so what does that tell you !
Because of not having a sweep go haywire I was not aware of the methods of checking and improving.
I see you are new to the forum by a few months and I have never had such attacks on my work before. Dont go overboard and have a go at me at being thick and crappy with my drawing as I am not. I am having to follow drawn dimensions and until now all went well.
I do EXACTLY what you are trying to do for a living
post an example of the sort of crappy plans I have to work with, as I doubt anyone is doing what I do !
Also I have found that V4 does not do that twisted thing as V5 did, so I was not aware until V5 of the potential issues following paper dimensions could give.
I am now taking on board the calm advice I am given, treating every curve I draw to the curvature graph, altering the shape that was supposed to be followed so as to keep Rhino and yourself and the forum and the sweeps happy.
The other day I established where the hinge line had to go, this then meant the existing curvGraph adjusted curves no longer went through the first plotted hinge line. Again I had to alter them. As I introduce objects from other plans where the plan drawer had not the benefit of CAD I have to alter existing curves to make things fit and that can see originally good curves get degraded. Now I have seen such trigger a bad sweep (note V4 hid the badness somehow) I have asked and received advice and am the better for it. Manuals most welcome but the forum is also here to help folk fine tune their work.
I am also making for an environment where accuracy is not paramount so having Rhino throw a wobbly on curves where my client looking at them would say its good enough is also a burden I have to take on board with more time needed than client wishes to expend.
I really do look forward to following the Level 1 and 2 pdf.s and very grateful to at last see these as I have asked of such in the past and not seen them.
you’re thinking that you’ve found bugs
I do not say I have found a bug, I ask of what is happening, sometimes it is declared a bug, Pascal declared so recently, then more recent someone else. dont have a go at me for declaring bugs for anything that doesnt seem to work, I am aware that V5 is different. I spotted a problem with dimensions, something that didnt happen in V4, its a designed in sort of bug and I now work with the workaround offered by Pascal.
I’d rather saw my left arm off than help you at this point.
I must think myself lucky that amongst your response you have given me some help with the two links,
Of all the posts I have made over the years, I have never come under attack like this before from anyone.
Steve
But there is no one like You bombarding the forum with endless messy questions!!!
No need to be personal here folks!
Anybody can use the forum for what they like.
The questions asked are GOOD questions. This is what Rhino CAN do with the input curves:
I just copied the base curve to the top with Orient2Points and used NetworkSrf.
So the question to ask is: Should more commands have the ability to act like NetworkSrf? With the ability to rebuild all inputcurves based on tolerance?
So @Steve1 see if NetworkSrf can be your tool of choice. I imagine you need to have the fourth curve anyway, since you are focused on high accuracy.
Thanks Holo for some defence, forum has turned a bit uggly. If most of my questions are messy then the mediator needs to intervene. Likewise on unwarranted comments.
I thought my questions were ok, any messy ones, and there must be loads to choose from ref bombardment, please indicate examples and I shall refrain and go without ever knowing the answer.
likewise others wont get to know.
Thread on twisted sweep proving useful with 260+ hits , if its a messy unwarranted question then 260 wont learn anything.
Holo…I couldnt use the top rail as the data had nothing there for me to refer to, I was using the sweep to help establish what might be there, an overhanging object was to be projected to the skin, then that curve adjusted and refined, then another sweep made with then the four curves and sweep or networkSrf used, I try both as well as sweep using top/bott for rails then change to left/right for rails to see which fairs best.
Steve
Steve1 -
Fair enough, here’s an outline of the work I’ve done that I think is applicable to what you are trying to do:
I started designing scale RC aircraft in about 2004. This was done using AutoCAD, which I started learning in about 2002. I have my own laser cutter and made and sold my own kits. All of these planes were designed from one of two sources - either old 3-view drawings, or converting existing model airplane plans into CAD and having to make things actually work and fit. Here is a fairly complete list of the planes I made and kitted, these are all links to the build threads I did on them:
http://www.rcgroups.com/forums/showthread.php?t=416630
http://www.rcgroups.com/forums/showthread.php?t=527598
http://www.rcgroups.com/forums/showthread.php?t=444771
http://www.rcgroups.com/forums/showthread.php?t=469165
http://www.rcgroups.com/forums/showthread.php?t=469165
http://www.rcgroups.com/forums/showthread.php?t=537322
http://www.rcgroups.com/forums/showthread.php?t=662704
http://www.rcgroups.com/forums/showthread.php?t=662704
http://www.rcgroups.com/forums/showthread.php?t=480576
After doing model airplane kits, in around 2008 or so I transitioned to working for a shop that specialized in experimental aviation. It was at this time that I started using Rhino, and I started getting into CAD/CAM and digitization. I purchased a FARO arm used off ebay, and started using it to digitize existing aircraft structures, then use Rhino to design modifications to those structures. I then machined the molds for those add mods, did the fiberglass/carbon layup work and installed those parts. Here are some links to my blog/website that show some of the projects I did while I was there:
I started learning T-Splines around this time, which I found quite useful. If you’re not familiar with it, T-Splines is sort of like NURBS on steroids. I was the first person to figure out how to maintain accuracy to airfoil shapes in a T-Splines model. Because of this, I was asked by T-Splines to do some webinars to show my process and share some of what I had figured out. I think in all I did about 5 webinars for them, but the one that is really aircraft specific is this one:
Also - you may have seen this section of my website before, where I talk about redrawing airfoil ordinate files to make them into nice NURBS curves. If not, it’s directly applicable to what you are doing and you might want to read it:
Around 2011 I struck out on my own and also purchased a laser scanning setup and photogrammetry setup. I run this business with my partner James and we keep very busy and love what we do. Here is a link to our site:
http://schoolstreetdesign.com/
The vast majority of the work that I’ve done for the past ~3 years is wrapped up under NDA. I can tell you that about 50% of our business involves making NURBS models that are used for CFD analysis in the aerospace industry. The source data for these models comes from a variety of sources, such as:
Old blueprints
Old point cloud data
Old CFD Mesh models
Existing NURBS models
Existing 3-view data
ALL of this data has one thing in common. It’s “noisy”. I can tell you that even 3-views drawn up in CAD taken right from the Boeing website will have discrepancies in them. ALL of these data sources create the exact kind of headaches you’re experiencing right now. Dimensions are off. Drawings are distorted. Sometimes the data is just plain wrong. We do about 30 of these models a year.
We also do full NURBS models of aircraft from laser scanned data when there is no existing geometry or documentation. This spring I did a full laser scan and surface model of a Lear 60. My customer showed it to an engineer who worked at Lear when the 60 was developed, and declared that my surface model was actually better than the one they made the plane from. I can tell you that even laser scanned data of the actual aircraft creates many of the problems that you are struggling with. Really, all data has a certain degree of noise, and it’s our job as the surface modeler to extract the essence of a shape and not get caught up in the minutia.
Often these models are used for analysis for proposed changes to the exterior shape of the aircraft. So, say if someone wants to add a large antenna to the top of an Airbus so the folks can get internet access in flight, then someone needs to make sure that the overall performance and stability of the aircraft is not affected. So, this is the type of work where lives are literally on the line.
I also do quite a lot of “radome” work where I actually design the fairings that cover these antennas. For these jobs I typically get a set of cross section curves from an aerodynamicist that my surface must fit to within a tolerance. Again, often times these input curves are messy or not conducive to making nice surfaces. What the customer is looking for is something that fits their input curves, but is also smooth and free of any defects so that molds can be machined with a minimum of secondary work. This seems like the very balancing act you’re having trouble with. I can tell you it’s a big thrill to board a flight, look down the fuselage and see the surface that I made sitting on top of the plane!
Sorry for the delay in responding, I was laser scanning an aircraft all day yesterday. If you think my background qualifies me to give you advice please let me know. I’m booked up the rest of the week with paid work on short deadlines, but should have some time next week. I think what you need is someone to give you a quite long explanation with a lot of context, but I’m not going to spend the time doing that if you don’t think I’m qualified.
-Schuyler Greenawalt
But look at that surface!
That is definetly not a desirable result.
Granted it’s better than the result in the opening post, but surely that surface doesn’t have a lack of control points…
Norbert
I think what ticked off some people, was that you came across as somewhat reluctant to take the advice to learn the basics and clean off your base curves.
At the same time the multitude of new threads you posted in a very short time didn’t help to straighten out that impression.
Cheers, Norbert
That is not for us to decide IMO, nor is it the point as you surely understood.
If the user needs 0.01 mm tolerance from the initial curves then this is close to what the result will be no matter what tool you use. If you fit new curves to the existing with 0.01 tolerance then they will have a LOT of controlpoints, and NetworkSrf can be sat to have 1 or 10 untis tolerance and create quite simple surfaces.
The point is that Rhino CAN use the curves and generate good geometry. And I would like to see that on other tools as well.
Sometimes we need to work with joined, trimmed edges, and they can be super-packed with control points. And this is one of the things that make NetworkSrf a great tool. But I agree that any good modeler needs to understand what is required for the tools to operate as planned.