Following the recent video from @theoutside and our discussion on the related thread, I’m asking for advice, or explanations about what I’m doing wrong here attempting to rebuild this thing in Nurbs, going for a slightly different design compared to the original.
I have made relatively clean boundary curves.
And a single-span extrusion that I would like the recess on the side to follow.
Attempt 3.
Project the front curve on the extrusion.
First this is problematic because the result has a lot of CPs.
So I use a loose projection of the front curve on a loose offset of the extrusion as my reference curve.
So now I have this.
This has crossed my mind : intersect with history and move control points of the sweep until I get close enough to the target curve. At this point I might as well switch entirely to SubD…
Bonus problem
This is an attempt to model the top notch… A problem for the future.
Yes, I know I can do this in SubD. SubD is great but it cannot be the default solution as soon as Nurbs seems overwhelming.
Yes, I know I shouldn’t try be so close to the image, that’s just a “reference” as Kyle says. But let’s assume I have made those curves from scratch and I’m happy with them. Now it’s surfacing time, the issue is the same. Or I may be in a position where I need to edit a piece of a model, and I must attach to boundary curves.
Yes, my brain is not made for freeform surfacing, each time I hear someone doing a scale instead of an offset with a fixed distance, I want to throw my compurter by the window
But I’m interested in learning this workflow.
Should I quit right now and stick with Grasshopper, or is there a nice, clean, methodic, intelligent way to approach these surfacing problems ? I could add a huge list of 5 sided-patchs to the menu since that’s also a situation I find myself in quite often.
I’m attaching the files if it’s useful but beware these are ongoing attempts, totally messy. I think it would be better to start from scratch.
this is a very common problem with organic models and often requires some iteration to get it right.
I’ll often project curves with history in both top and side views and adjust both until I get something that is close. Then I’ll often trim out bits and pieces of both curves and blend them together in order to get a shape I like on the surface. Or sometimes you have to actually adjust the surface in order to get the projected curves to lay out right.
now… there are times this won’t work because your surface shape is just wrong and you are making an impossible transition… take the time to really evaluate it from all angles and adjust the surface… with history, your original projected curves should update and you can see the results in real time.
Well the underlying dilemma here I see, is the matter of ‘reverse engineering’ a component from one single purview or perspective.
It’s fine to attempt this, but it should be considered to regard what on Earth the design intents actually are or were behind the object in the given photo.
It’s highly possible that the said object is obsolete or easily outmoded by anew version built from scratch and total freedom of new design.
However, not knowing the strengths and weaknesses of a competitors design, will make it difficult to come up with a better or even similar version, etc.
Ultimately, it’s not worth the effort, stress, or mental capitol to go down an endless tangent of assimilating technology that is already obsolete or antiquated by more modern technology. So it’s quite possible that the product in question here isn’t the one that would make the most use of anyones’ particular time etc.
In terms of rotary tool technology, if you’re not focused on the type that is ‘brushless micro motors’ then you’re wasting time on assimilating ancient technology. But is that’s what you want to do that’s of course you’re prerogative.
If you go brushless, you wont have these silly brush port geometries:
Now, obviously this may have more to do with an inquiry for particular modeling techniques of NURBS in general – which leads me to the question, what’s your question here regardless of the geometry you’re trying to assimilate?
Because, maybe the geometry is just wrong to begin with? Or what are the shapes you’re trying to model without sub-D?
Cause, anything is possible in Rhino. So, please make your questions more clear. I don’t think it’s really about the object in this photo at all. But I could be wrong.
An archeologist could spend decades trying to figure out the design intent here, with multiple versions or permutations superimposed here to reflect infinitely many possible directions one could go from here:
Yes, you are right. That was my first inquiry here.
And the answer was to post a specific file, which I perfectly understand as it’s the standard on the forum to solve specific issues. So I took this recent model as an example but really I have dozens of similar files with similar problems I can’t solve.
On another thread you said :
And I couldn’t agree more. I’m in a desperate need of training material on how to approach modeling - that is, really, before even starting to draw a single curve in Rhino. I know the tools and their options. What I’m missing are more generic ‘skills’ that are in fact not Rhino-dependent.
Problem A
How do you “read” a picture or even a real object you want to reproduce ? How do you identify if the model is made of a single surface, a bunch of patches, where are the seams, how are you going to hide them : simple fillets, blends, more complex transitions ? What are the clues, the characteristics that can give an indication of how flat/bumpy/twisted/whatever a surface is, where it starts and where it ends ? This is especially hard when there are no “hard lines” on the model or there are disappearing edges.
And let’s clarify now that my job is not to design things. So I don’t care to copy existing designs, as bad as you may think they are, this is entirely for training purposes. I also think analysing how existing things were made is a necessary step anyway to get a deep understanding of the tools and methods that allow you to be creative. Maybe I’m wrong.
In the case of Rotary, yes, there is only a single picture. That’s often the case though and it’s not even always an orthographic one. The question is : how do I read this picture to get some information that I can use to build something that respects this front view, with some liberty in the perpendicular axis ?
In this case I think I can use the outer boundary and the inner boundary. Let’s say I want to respect those two front curves exactly - I like them, I need them, whatever.
Now the question is : given these boundaries, how do I fill the gaps ? Is there something I miss ? If so, can I extract it from what I have ? Or is there nothing more to see, in which case I’ll have to invent it, fine, but still I would have to know what I’m missing. Because with only these curves, I have no idea how to patch this correctly.
Problem B
And then there is the topic of surfaces with more than 4 edges. I know, Nurbs doesn’t allow that, and yet I have at least one of those situations in every model I’m attempting. So again, the specific questions would be :
1/ am I doing something wrong repeatedly and can those be avoided ?
2/ if you decide to use a four-sided patch with a trimmed edge, how do you know which of the 5 edges in the trimmed one ? Is there a way to match a trimmed edge to an edge ? (I know refit or point editing until the shape looks close enough but both methods are not exact.)
3/ if you decide to break the patch in smaller 4-sided patches : what characteristics, what indicators can you use to choose where to divide the patch ?
4/ And then, what is the proper matching sequence, are there, again, rules that have to be followed so that the process is actually smart, and not an infinite random sequence of matchSrf - changeDegree - InsertKnot until miraculously everything is G1/2/whatever you need ?
See these other threads for examples.
This last one is an old thread but I encountered it again recently and tried many things… including doing it in Catia. I’m out of ideas. SignetRing.3dm (2.4 MB)
So, summary
I’m looking for generic rules, checkpoints, pitfalls to avoid, options available, so I can at least have an indication I’m on the right path when modeling. Because it’s so easy to spend hours trying to build something that is anyway impossible and not realize it.
One approach is to envision the shape in 3D, typically with the aid of additional information. In another thread Kyle said
Kyle may have started with an initial vision of what the rotary tool should look like based by a physical example, then created the sketch based on that vision. When he did the 3D modeling he presumably had the sketch to work from but also his vision of what the object should look like in 3D.
Also in these videos Kyle is generally creating a 3D concept model with few constraints.
Look at simple objects around you and think about how they may have been designed. Learn to initially ignore fillets and other transitions and concentrate on larger surfaces. I frequently try to guess how the object was originally designed and/or made.
Kitchen utensils and tools can be good examples. Pick something you have which is simple and small enough to set by your computer. Then try modeling it. Initially don’t worry about exactly matching dimensions or shape. Instead concentrate on how to structure the model to represent the object.
I’d be interested to see samples of these challenges.
This is familiar to me, when I first started using Rhino. At the time I did have a background in many other CAD’s already though, but Rhino is quite unique compared to many other CAD’s.
Rhino is indeed ‘freeform’, which can have it’s pros and cons. But Rhino is evolving so much that ‘freeform’ is also becoming evermore ‘parametric’ – which blows the mind a bit.
When I really started diving into Rhino back in 2008ish, I printed off the whole command listed and just started reading it over and over and over.
As I did that I began recognizing many commands as ones I’ve already become familiar with in other CAD’s, so this helped me tremendously in my beginnings.
See, as I saw it there were basically at least 3 or so ways to executed any feature in Rhino, by using either the dropdown menus, command prompt, or clicking the tool pallet buttons, etc.
So I figured the first way I’d become proficient is using the command prompt for practically everything.
Yes my boss would tease me as he looked over my shoulder watching me typing everything like I was some kinda hacker in a hollywood movie or something lol.
But that’s what I decided to do from the start. I’d already been familiar with AutoCad’s GUI with the command prompt, so I wanted to see how far I could go with it.
Over time yes I also became familiar with all of the dropdown menus, and pretty much all of the tool pallet buttons.
I always had fun displaying all of Rhino’s tool pallet buttons on the screen and just browsing through all of them for hours there’s so many. That was back in V4 and maybe V5…
I honestly haven’t done that for V7 yet, I probably should I did kinda skip V6 a little, I felt like V5 was better at the time cause I couldn’t get my colleagues to convert to the new one yet. I’m still trying to push them into V7. When V8 gets released it will be fun
I digress, obviously ( skills in fact not Rhino-dependent ) reminds me of how most ‘windows based’ programs lets say, have the whole ‘dropdown menu thing’ and maybe ‘tool buttons’ thing, but the ‘command prompt’ thing is sort of unique, well maybe not so unique considering DOS or CMD etc., and there are programs that kinda have a command prompt for different things.
But I guess I believe that Rhino’s command prompt is quite unique in comparison the many other command prompts I’ve ever used.
Moving on though, I think yes there is a dire need to address the whole ( not any particular CAD-dependent ) approach to designing anything new in the world that needs to be created that never was.
This might make due for anew thread if there aren’t any or many yet, or certainly may continue this one on this matter, and find others – however much similar. Maybe include all of them, or compile them etc.
Very excellent questions. The answer lies in the ‘design intent’. Yes that’s my short answer, cause this can go different ways for every single design, and it ultimately depends on the decisions of the ‘user’, and how they want to achieve what the need to accomplish and the compromises that may need to take place. Cause there’s most likely going to be compromises in any design.
I think you’re correct. The difficult part though is knowing the details that are behind the scenes of any design, and how it was achieved – this is not always obvious at first glance, and may need deeper reveal of the geometry in question.
Yes, these things are entirely up to you as the user/designer. The concern I have personally is, how much mental capital will ‘I’ as a ‘user/designer’ invest in jumping to conclusions on some particular image of some particular component…
Ultimately, I think it all depends on how you’re trying to capture some particular object/component.
Obviously if there’s only one particular image, then it’s quite impossible to capture much of it very well.
So, right away the design tolerance to the original is like +/-0.5" – which is quite large of a deviation.
This deviation has it’s pros and cons. I think not worrying about accuracy here in cases like this should be embraced as a benefit, and the imagination should be allowed to run free.
While this makes it difficult to answer a few of your questions, cause any particular phenomena has and infinite number of interpretations.
But I think you demonstrated some of this here
This reminds me of my own journey with the paths of reverse engineering. It’s as if you’re identifying the questions, and the answers on your own, which is leading you to the next steps in the workflow.
The next steps in the fork in the road. Like the next part of an algorithm to a recipe.
Ut oh yes the proverbial trimmed srfs and multi-sided srfs
It depends if you’re reaching the end goal or not? Ultimately, I think we’re all supposed to be creating the world that never was, so that it can be a better place. Therefore, we’re supposed to be creating things. Yes, there’s many trade-secrets on how to do so at every step in the wave-line from beginning to end of every recipe.
So the main question is, did you make that which you set out to create? or did something happen that prevented the creation of said attempt?
There’s a few ways to do this, my favorite one is to set a specific color for ‘naked edges’ in a particular display mode. But another way is to run the ‘show edges’ tool and display the ‘naked edges’ that way.
No not really. That’s why I don’t like ‘trimmed’ srfs. There are alternate arguments of course, but those just avoid the obvious which is ‘untrimmed’ is better than ‘trimmed’ – don’t let the ‘trimmed’ members fool you.
But yes, ‘trimmed’ srfs are kinda inevitable. The most common ones are the ‘planar’ ones that aren’t 4 sided, 3 sided, or 2 sided, and/or have holes in them.
Not sure what you mean by ‘divide patch’, like number of UV isocurves?
There’s certainly a few tricks. The nomenclature of ‘work flow’, comes to mind.
Bending the rules is where I tend to lean. And maybe break them if necessary.
I used to do the whole ‘insert knot’ thing, but I haven’t done that for many yrs.
‘network srf’ is the ultimate tool for everything imo. Some ppl will argue against that, which is silly imo.
Every single srf … has a network of curves that comprise it. period.
If you can learn to ‘wield’ the networks that comprise surfaces, then there are no limits.
It would be great if NURBS could be different shapes other that UV though
Very interesting. I’ll take a look when I can.
Indeed, this is why I usually ask questions in my mind like this one for example:
Is this object that this person want’s to design, the shapes they want them to be or are they random, arbitrary etc…?
I’ve also encountered many designs where the shapes are the way they are, due to the limitations of the designer that “designed” them.
You never know when something is shaped the way it is, because of some limitation of the ‘user’ that shaped those shapes.
With that being said, I do my best to be careful not to tarnish the designs I see from day to day, but in recent times I’ve began to assume nothing is exactly what ppl want them to be.
So, from the standpoint of a reverse engineer, there’s no need to capture things 100% as they are, because it seems most ppl don’t really know how they want their designs to be exactly.
And the world has a lot of flexibility in designs and functions.
New users that don’t know what changes as they go forward, will encounter many alterations that seemingly don’t make sense. So rule 9 there, even though it’s a different program, definitely points out some good things to think about in terms of workflow.
But still I think there’s room for compiling something more clear and concise.
Rhino’s help documentation, I’ve found does a good job of this. I’ve referenced it alot in my days.
I’ve also seen the help docs become somewhat more simplified in a few cases, but I’ll address that later if I can. I think the help docs should always be very thorough, and not leave anything out for the sake of laymans terms.
one rule for modeling things I designed myself is “the sketch is only a guideline, the model is the rule”
I used my own sketches for my own designs only as a reference to get the model started. Once I am in 3d I make all my design decisions in 3d and only use the sketch to make sure I’m sticking with my original design intent. The fact is you can lie, cheat and steal in a sketch and there are no consequences …in 3d all those cheats come to full light and you have to deal with them.
Unless you are trying to exactly duplicate something that already exists, you should allow yourself the freedom explore and improve your designs by making your final decisions in 3d and not stress about the sketch so much.
I did kinda skip V6 a little, I felt like V5 was better at the time cause I couldn’t get my colleagues to convert to the new one yet. I’m still trying to push them into V7. 
