Surfacing... just a nightmare

Sadly, Rhino lacks a static Zebra analysis, which could speed-up the fine adjustment by a lot.

My favourite part of the VSR plug-in for Rhino 5 was the extremely good coloured stripes of the Light lines analysis that it introduced. Their colours were cleverly selected and definitely do a better job than the similar analysis tool in Alias.

Hi -

Have you tried the GHGL test that was posted here:

I’ve now made sure that this feature request is on the list of things to track - RH-58682
-wim

1 Like

I don’t have any idea how to use it :smiley:

I could not agree more. I think NetworkSrf, like Patch should be used almost never. NetworkSrf should literally never be used to make primary surfaces. Ever.

3 Likes

Oh dear god…Steve1 is that you?

Steve1 is that you?

Huh?
Thats only 10th of mm in the real world.

To clarify, I am calculating Tangency without iteration, so its an approximation. Using 5dec places results in accuracy to about 3.

f

When I loosen that to .00001 it results in 1

g

There are no tools available for doing this in Rhino or GH.

many of the smaller ones do- and many of the larger ones do yet don’t talk about it… :wink:

Okay let me explain the Steve1 reference, because it’s actually informative. A few years back, a fella by the name of Steve1 came on here looking for help making models. His project was super duper secret (obviously making WWII models, likely for making scale models, but like shhhhhhhh!). We tried to help Steve1, we really truly did, lots of us. He was struggling with the absolute basics - getting things watertight, patch layout, lots of general “how do I make this?” questions, which almost always involved for him feeding a giant rats nest of curves into NetworkSrf. He seemed to have lots of preconceived notions about how freeform surface modeling SHOULD work, and then ranted about how things didn’t work the way he thought they should, while clearly not ever understanding how things ACTUALLY do work.

You’re not doing yourself any favors by being all cagey about what you are actually trying to model. If you want help from the general community, start with “hey guys, this is the shape I’m trying to model.” You will get far more instructive answers, and people can likely steer you in more productive directions. This whole “let me show you this ultra close up of two edges that aren’t watertight” doesn’t actually help anyone help you - to help you solve that problem, people need to understand the greater context of the surfaces around that naked edge. If that model is somehow under NDA and you can’t show it, but you’re struggling with issues like simply making your model watertight, well, god help you and Steve1.

Don’t be Steve1. If you want help modeling something, start at square one. “Hey guys, I’m trying to model this type of shape:” The first thing that you need to consider on ANY model is patch layout. Literally, that’s square one. If you’re not showing your overall model, no one can check your patch layout. Everything that you post seems to indicate that you are in way over your head, so if I had to guess then I’m betting many of your problems start at square one - patch layout. You’re answering questions by zooming in on minutiae, which is helping no one, least of all you.

Seriously, this is so Steve1…

14 Likes

What have I done wrong this time?
Poor me, always at the gallows … :zipper_mouth_face: :no_mouth: :upside_down_face:

I would lie if I would say it didn’t cross my mind too.
Might say more about us than about Proterio though…

1 Like

Im dont want help modeling, and I dont want my patch layout checked thankyou. I had my physical recently ) … If you are interested in solving specific problems, I have posted 2 so far, Tangency & Split that remain unsolved.

I understand where you are coming from. My project isnt super secret. Some of the people replying to this thread know all about it. I also understand that Engineers just want to tear into the specifics of the project and solve it. If you want a shape, picture a bathtub, yes seriously, and the problem of calculating how much water it holds. That should not be this difficult.

As i said thats like giving me a fish. I eat for ONE day. I am asking more general questions in this thread so I can learn to fish. I have had to defend why I am using a tight tolerance, and why my curves dont line up. I did. I am capable of modelling this once I grasp whats required to make surfaces match.

Diving into this specific challenge is just going to change the focus to the specifics of the model. There are many threads that do that. This one is a more general discussion, with your permission )

I am willing to bet many people have faced these challeneges modelling. In my discussions with people they all they say the same thing - practice for 6mo then you will “get it”. Most are not willing to be this pushy and procative. I do so to expand knowledge for more than just myself.

I have watched a lot of videos, read a lot, hired experts - some that posted in this thread, worked on this for months and STILL surfacing results I desire elude me. Is this a problem with me failing in the ways you suggest, or is it that surfacing needs a LOT more discussion from a first principals perspective?

I am not sure what you kmean by patch layout. Perhaps you could post a link to a video or thread… that way those that follow can benefit also.

As for being way in over my head, Ill ignore the condescension and remind you that all experts started where I am. I get that this thread is irritating to you, but some surprising and illuminating answers are emerging.

To use your fishing analogy - you’re not fishing, and you’re not making it any easier for anyone to teach you how to fish. You’ve disassembled your reel and you’re staring at all the internals under a microscope and then wondering why no fish have jumped into your boat. I know you think I’m being condescending, but I, and lots of other people on here are trying to teach you how to fish, not simply giving you fish.

You’re really making my point for me here. I started where you did. So did many other people on this thread. I haven’t lost sight of that, but I’m telling you, straight up, you’re making it impossible for people to ACTUALLY teach you how to fish. Start with “This is what I’m trying to model” not calculating tangency to five decimal places.

3 Likes

OK enough with the Holier than thou criticism. I got your point. Post some links or move on. Thx

1 Like

If you’ve gone through the Level 1 and Level 2 course material provided by McNeel, then the first ~150 pages of that tutorial is the best thing out there to show you how think like a surface modeler.

4 Likes

@Proterio you didn’t give an example of a file where the tangency fails. btw: I think you’ve got your unit settings wrong. If you need 0.00001 tolerances in meters and say but that’s only 1/10th of a mm (it’s 1/100th btw), it doesn’t matter. Units are units to Rhino, no matter how you call them. So better model your stuff in mm so that you scale your model by a factor 1000 and use a normal tolerance like 0.01-0,0001

1 Like

I looked at that Audi R8 Project. 600 detailed pages is too much of a deep dive. Everyone is selling something and still no discussion of the concepts that seem to flaw my attempts.

As I mentioned, I watched 100s of hours of Youtube videos like this

These “monkey see, monkey do” are not helpful when it comes time to design soemthing unlike the examples. I know how to operate the interface. I have completed all the Level 1 and level 2 training that pertains to my desing work:

There are these models

but again no discussion of the approach and why that approach

3DMax is a common choice for car modelling it seems. I am not a fan of polygon modelling, probably because I dont understand how you can generate a smooth curve over a long span without splines.

I gave up modelling a car body to work on something more simple because I dont believe that 600 pages will reveal the undelying concepts that cause surfaces not to join or merge!

Contrast that to this response:

So better model your stuff in mm so that you scale your model by a factor 1000 and use a normal tolerance like 0.01-0,0001

Now thats what Im talking about. That is illuminating and a conceptual approach based on years of facing real world challenges. Its one of those… duh… of course. Why-didnt-I-think-of-that suggestions.

Thats what this thread is about.

Not so much fails, as not really tangent. This is a description of the challenge

This is the problem with the GH module

Can I propose we flip your question on its head? Why are you approaching modelling in the way you are? As far as I can tell you are using Grasshopper as a panacea to reduce rework of a model? Is this model being used for manufacture? If so, it’d be handy to know how the object is being manufactured; eg, a tool for a rotomould does not need as fine a tangency tolerance as a polished injection moulding tool. Once you provide information such as this then the answers may be more focused, as your quest is fairly nebulous at the moment.

One thing I would add to the mix - edge surface. This is the main way I create surfaces that have a fairly even UV length ratio. This is because it respects the degree and CV point count of the input curves when creating the surface. As others have mentioned, use either smaller and multiple surfaces to capture a complex form instead of expecting a high degree single span surface to do the same. Or use lower degree multispan curves and surfaces, as you can increase the CV count and also benefit from the localised curve change present in a multispan curve/surface.

The problem with almost any car tutorials on youtube or ebooks is, that they are made from people never worked in the industry. They simply teach the wrong stuff, doing more wrong than good for your understanding.
The only valuable tutorial source shown in this thread are the videos created by Raymundo Burgueno using Alias. He is more explaining about the why and what and not about the how.
You cannot fully teach yourself or others modelling cars or other complex shapes, if you never seen a design studio from within. But at least you can tell that a 16 years old car modelling expert using Blender is not a trusted source of truth :slight_smile:
In times of google its really difficult to filter the right information, but at least you get some information nowadays.
So the only solution to become better in surface modelling is doing it again and again. Listen to the right people, and ask for the why. Ask questions like: Why are twisted surfaces bad? Why do I need to curve any surface. Why should’t I match on trimmed shapes, Why using static and not dynamic highlights and why do I need a curvature graph? What is it telling me? Why did software developer developed this tool? You see, if you understand what the purpose of many things are, then it really doesn’t matter which tool to choose. You can reach many things on different ways, with different software packages.

Really learn from the space ship tutorial shown above.

1 Like

Can I propose we flip your question on its head?

Sure. Why not.

Grasshopper as a panacea to reduce rework of a model?

Yes, mostly. I am not bright enough to be able to remember the path through the maze 30 days ago, or even 2 days for that matter. But GH is a parametric tool. Once I have summited this mountain, there are more. a structure must be built to stiffen this “bathtub”, and once that is done, we have to see if what remains is big enough. If not, some sliders will be moved.

it’d be handy to know how the object is being manufactured

NOT injection moulded, welded sheet metal of some sort probably, so next is a thread on panneling challenges. Brace yourselves )

your quest is fairly nebulous at the moment.

Well yes, and deliberately so. Diving into specifics doesnt really help the big picture. Like for example this little chiken and egg surfacing problem:

h

The intersection of these two surfaces produces a curve. That curve is the tangent transition along the two surfaces BUT, the red surface was created without a vertical curvature that the finished surface will have, so the curve will change. The red surface is a lofted surface from two curves. It needs three curves or more, but to make a curve in between the intersection of these two surfaces is required… another iteration problem that must be approximated somehow.

If I use the curve generated above, then when the red surface is lofted using the now completed set of vertical curves, it will have an artifact resulting from the innacuracy.

I dont even know if its possible to curve the red surface using a curve, an arc in the vertical direction. My current idea is to use NETWORK SURFACE and thus include the arc to derive that curve (which will be different)

Regardless of the model, the approach you take to modeling is crucial. I began thinking that I could just trim away and join surfaces to get a perfect result. I have abandoned that approach. I now use the trimmed surfaces to generate curves from which to loft one surface that forms the entire side of the bathtub excluding the bottom which I hope to join with a fillet along the length.