Surfacing... just a nightmare

Yes I know the problems, Its the solutions that we are discussing…
You guys are making my point

Did that. Now im here. I backed away from the car body because that was clearly the high bar. Now im working on a very simple bathtub shape and I find myself reworking and reworking my approach to navigate all the pitfalls mentioned.

The solution to the first classic loft problem is this:

Notice the TWO CURVES AT THE FRONT THAT DO NOT TOUCH yet make a perfect underside surface… (no discussion about that nifty little trick I note)

k

Only minor surface anomolies

LoftTechnique.gh (7.9 KB)

Then there is this approach with a surface-

Also discussed here:
Boat hulls with smoothly curve stems - how to model?

The problem is that you enter TSPLINES territory and the inherant problems really fast (ask Kyle, THE TSplines expert) . (also not discussed)

So where is Mr. close the thread who wants a problem to solve?

Start with the Loft of three curves I posted, all identical and create a nice smooth surface like the one shown above and DESCRIBE HOW YA DID IT !

HOW do you chop of part of a curve such that the surface is not ugly at the curve end and produces a surface with degree 3 isocurves. Then we can move on the more complex problems like this:

A brilliant solution, but difficult to merge into surrounding surfaces. This needs a geometric solution.

This is thread about the APPROACH to modeling complex problems like this.

Static Highlights: It has already been answered. Zebras are misused as a continuity analysis, especially in Rhino. But its not the best way of judging it. Its rather about seeing the reflections, its flow and its proportions. Furthermore static highlights better reflect the lighting situation in a real world szenario.
If you want to judge a moving light, its a better to use some sort of directed point light, similar to a flashlight.

Twisted Surfaces: One criteria of a good model is to have very controlled curvature. You want it proportional and as homogenous as possible . You also want your edges to be of similar character which means they only diverge minimal. Of course you sometimes need to create twisted surfaces but this should be rare. The problem with twists is, that they change the reflection of your surface, making it very small at one end and very big on another and may also create local flatness or infliction, but even worse: If you match a surface to a twisted one, you also get a twisted one, but now you‘ve got a sine profile which is extremely bad. On a car this would look like a fender bender.

Grasshopper: To say it with HAL’s voice: Don‘t do this Dave. It is the absolute wrong tool for surfacing. It can be used as a tradeoff for surface pattern creation, but it is no replacement or improvement of surface modeling. It lacks too many tools. So do baby steps in Rhino and see what should and what can be automated.

1 Like

Wow. Now thats a statement and a half.
Then what? just Rhino and rebuild EVERYTHING a million times?

Yes! Don‘t expect one-click solutions. There is a primitive History in Rhino and Alias. Parametric surfacing is the domain of parametric modellers like Catia, but they don‘t do car skins/class A/freeform with that. Its more the domain of technical parts. The reason for direct modeling is primary the ability of controlpoint modification. You can‘t do this efficiently parametric. There is one hybrid called the ISD module for Catia which does that to some extend. But its expensive and quite a new technology.

Experience and light models help you in reducing the effort of rebuilding stuff again and again. The better you get, the better your first guesses and the faster you are. A skilled designer can be really really fast. If you don‘t know your form, make a sketch beforehand.

Its a bathtub with cylinders for corners on all the faces, two sides, front, back and bottom and some other “surface features”.

You obviously know a lot about this. Im not expecting one click surface. I just want an approach I can actually use. If not GH, Catia, etc etc what parametric modeller DO you reccommend? I cant do this manually in Rhino. There are too many unknowns. Its got to be parametric. Have you looked at MaxSurf?

Skylar, Just want to thank you for this litte treat - SCULPT & MATCH - a reel fishing lesson ))
(FYI, you siad “right” 453 times)

“we dont want to use trims unless we absolutely have to” - Fishing NEWSFLASH !!!

Pull does a lot of things PROJECT does, but better - NEWSFLASH !

Staring at a Zebra is crazy… use AUTODESK SHAPE ANALYSIS (no longer available) - NEWSFLASH !

The quicker you get into surfacing, the easier to make changes - now thats an approach.

Just FYI, that hull can only be poured out of a can. Those curves are too complex for steel plate. Its not as easy when you have curvature and panneling limitations. That requires contstant radius curves to make convex/concave forms.

Hard to tell. If you are looking for a hobby CAD then maybe Fusion360 is the right tool for you. It is parametric and has T-Splines (Sub-D) surfaces.
Solidworks is also nice but expensive. Catia and Creo are great but unaffordable for most private users. More of them I don’t know.

From what I understand, Solidworks is not surfacing friendly either. Its more of mechanical design tool.

Well I don’t know what you understand under “Surfacing”. Again “Direct” Modelling vs “Parametric” Modelling does not say the one can do more. Its a different concept for different purposes. Freeform quality surfacing is a process you A. don’t learn in some days even weeks and B. is something you can’t und want to do parametric or generative. Rhino is a Direct Modelling software with a broad spectrum of features. Its a generalist software, and so its not perfect for specialised purposes. Class A modelling is primarily done in Alias or Icem Surf (both direct modellers). Alias offers associative modelling same as Rhino, which gives you some sort parametric behaviour but without the typical treeview.

Fusion, Solidworks etc. are parametric modellers which allows you to do more technical parts, parts where the visual quality is secondary. They also provide tools to create visual appealing surfaces, but you have no chance to create the same quality with those tools.
It seems like you are unclear about what you aim is. You should ask yourself what are planning to do. Furthermore if you are very experienced in Direct Modelling, it is easy to switch to parametric. The other way around is more challenging.

1 Like

Not sure how you got that impression. I am very clear about that. I am unclear about how to achieve large span, smooth surfaces in complex shapes as posed in the examples above. You suggest that GH is not the tool, yet the options dicussed are no better.

My mastery of GH is NOT the issue. Its the approach thats lacking - hence this thread.
Can you provide a solution to the lofting challenge?

GH doesn’t even let you match two surfaces together. So how will you ever create a nice surface model from it? You see I’m not questioning your understanding of GH. Its just not the right tool for general surfacing. You can create pattern and solve repetitive work with it. But it has limitations.
Actually I’m not sure what you plan to do. Its too generic to give you a concrete answer. But if you lacking a workflow, then its not the fault of other people or CAD developer. We can give you some hints. And the basic idea is to say, learn Rhino, learn Surfacing, understand the purposes of tools. Of course then you can use GH to do some stuff. I proposed to watch the tutorials from Raymundo (Handlebar3d) on youtube. He is an Automotive designer and tells you the right stuff. You can adapt so many things to Rhino eventhough he is explaining it in Alias. And if you want parametric features, then you need parametric modelling software, but then you must be aware that this comes also with a tradoff.

1 Like

With the greatest respect possible, I am tiring of this “go away and learn the basics” kind of reponse, or “why is that the question” or “if you use x tool, or y plug in you will have no problems surfacing”

The reality is, that if the guide curves are not single span, the patch layout not well designed the resulting mayhem is impossible to wrangle out of uglyness. If I leanred nothing else in this thread, the approach is what gets you to the summit, not stuck on vertical cliff face.

I posted the first challenge here:

Will you post a solution?

I proposed to watch the tutorials from Raymundo (Handlebar3d) on youtube.

Links please )

I posted one instructors solution, which is not bad, but not great. He designs curves so that the entire hull can be lofted end to end in one operation. Thats what I have been trying to do.

NOW I am exploring the concept of jumping from reworking curves to reworking surfaces as suggested by Sykylar. That will probably have to be done in Rhino, but the majority of the work in creating the PRIMARY surfaces (which is substantial) can still be done in GH.

This approach is new to me, and judging by his comment that he fixes a lot of models that come from professional shops using Solidworks. It seems like this approach is new to a lot of professional modellers too.

Intersting to me, is that he feels that SURFACE FROM EDGES generates the most faithful surface from the guide curves.

" A lot of people take all the curves for a hull and loft from end to end and thats just not the way to do it" - NEWSFLASH

and he only says “right” 233 times in this vid )

I would have told you that to begin with if you hadn’t wasted tons of time messing about calling a huge boat hull a bath tub. A boat hull IF modelled for production has to take production into mind and thus bending single sheets of metal is cruzial. But usually you design quickly with all tools available and then remodell based on production (Only way to get a good result). Thus patch layout is important. And set your tolerance to 1/10 of production tolerance. Here that is typically 1-2 mm, so 1/10 mm is good enough. It’s not a Nasa space shuttle so it will be welded together with the help of a big hammer.

Surface from curves, rebuilding key surfaces to minimum controlpoints and keeping things unrollable etc is important for a final model, but doesn’t mean crap for rendering or early sketches. There you just need 95% accurazy and a water tight model for volume calculations and 3d printing.
Networksrf is a great tool, BUT you need to know that crap in is crap out, and it doesn’t tell you that you feed it crap in.

One thing that will help you a lot is understanding how controlpoints affect a surface and how you can manually match them to surrounding surfaces. The easiest way of working with patch layout is if each patch has the same amount of control points where they match. Then just align them with the edge point and presto, tangency. (It’s basically what matchSrf does automatically)

3 Likes

I am now creating primary surfaces to merge together in the SCUPLT and MATCH approach I am trying now:

and the problem immediately becomes, creating good primary surfaces from good guide curves. Tangency is a thorny problem in this project as decribed here:

and a discontinuity of tangency with the guide curve will be visible in the surface.

Are you attempting to do parametric surfacing, a la Solidworks/Catia/Nx in Grasshopper? :sweat_smile:

I mean, Grasshopper is amazing, but that’s like using a fishing rod to pick apples from a tree… I’m sure you can pull it off, and it’s going to look impressive when you do it, but it’s not really the optimal tool for that job… (and there’s a high risk of getting a fishing hook stuck in your ear).

5 Likes

I think you got off on the wrong tangent there…
Gh is for creatin the GUIDE curves and PRIMARY surfaces in this new approach I am exploring now. The sculpting and matching will have to be done in Rhino (without a curvature or tangent tool like the one he relies upon)

The guide curves are not trivial in this design as detailed above. Ending up with a series of single span primary surfaces < D5 is not trivial either. as he says:
SCULPT LOW DEGREE SURFACES, MATCH HIGH DEGREE SURFACES

and FYI, I already have 4 fishing hooks in each ear as testified by the lenght of this thread, and im probably only half way home at this point.

This is how most create a hull in Solidworks or Rhino. The result is does NOT conform to guide curves. I have abandoned tht approach

Thanks for the tutorial Skylar, I don’t know why it didn’t seem to be appreciated from some of the comments. Just another way to skin the cat, and you have been skinning for some time now. I have just wrapped up a needlessly complex model using many of the same ‘surface from edges’ tools you share here and it has been a nightmare of sorts but from what I had to start with there wasn’t really any choice other than starting with a clean slate. I modeled to the outside of the molded hull and then offset that 9 mm. Pretty hard to do as a solid but even as the inner surface of the molded part I got a lot of naked edges that weren’t there on the outer surface. I found that I had to extend edges on the perimeter to cross the centerline and then fix the naked edges mostly through Sweep 2 and then trim the perimeter of the inner surface with the half hull plane. Then I was able to trim that plane with the inner hull and outer hull and get the 9mm surf which joined up cleanly to create a closed solid half hull with no naked edges. Only 93 surfaces in each hull half and I think only one patch. I could probably eliminate that patch now but I think that the CNC guys can mill a direct to female mold from this with no problems.

4 Likes

That actually worked great! A powerful tip. Thank you!

Naked edges (the dots) that prevented joining into a solid: