Isn't parametric modelling the wave of the future?

This was just asked in another thread:

…why in the world would you use an engineering tool to do industrial design work? There is a reason cars are NOT designed in an engineering application; subtle surface changes and definition of character and form are not the strong suit of engineering applications. Sure mechanical, drivetrain & chassis components (among many others) are “designed” in an engineering application, but the visual & aesthetic components? Curious to hear your responses.

I wanted to ask this question as well, only the other way around:

At the school I went to, we were taught parametric modelling from the beginning because A) it was the fastest way to get stuff 3D printing prototype ready and B) if the thing went to actual production, it was faster for engineers to do their part as well (I’m talking hairdryers and toasters here, though… not cars).

Now, I’ve been trying to teach myself to model in 3D on and off for over 10 years. I’ve tried most of the software packages out there one time or another; Maya, Max, Blender, Z-brush, Lightwave, Wings, you name it and I’ve never been able to do anything in any of them. Though I knew what a polygon was, normals, splines, patches and other sorts of terms you had to keep track of, or risk having errors in your model never stuck in my head. I just wanted to create. Why should I care about the technical side of it?

I also couldn’t seem to get it into my head how this all worked in 3D when the screen was in 2D. When you had a very complex wireframe on the screen looking like a spider’s web, it seemed that it was near impossible to see what you had to select. None of the modellers let me organize what I had created in any way which seemed logical to me (even though I was used to layers in Photoshop). Selections seemed to be in themselves as complex as the actual modelling, and if you did it wrong you could get holes, overlapping geometry or very, very sharp edges in your model.

But a whole new world opened up to me when my teacher told us about parametric modelling. It was wonderful, I didn’t have to care about keeping track of my vertices or anything. Every thing I did appeared clearly in a list for me to name so I could remember it and if I ever wanted to change anything, I could just look it up and change it on the fly. It was all completely logical and intuitive, not a technical thing to care about at all!

It was sort of like being in Illustrator rather than Photoshop to continue the metaphor. I could finally define a 3D curve using two 2D curves, which allowed perfect control that was again easy to understand. The selection problem simply didn’t exist, and at the worst, wrinkles would appear in the model if you did something “wrong” (usually though, you simply weren’t allowed to do anything wrong, and the application would warn you when you tried).

However, parametric modelling is mainly labelled as an engineers tool, something I only understand (probably out of ignorance) as this: 99% of the world’s 3D modellers come from the traditional 3D background, with vertices and splines and so on, and they know the technical aspects by heart so they are not bothered by it. But hasn’t that technical knowledge been made obsolete by now? One of the “new” ways to “traditionally” model in 3D is by subdivision surfaces, which does indeed to away with most traditional technical aspects.

So, why can’t parametric modelling be the new way for designers to work?

I found a video which seems to indicate that new techniques and breakthroughs in flexibility are being made in this area too. Just skip by the five or so minutes of sales pitches in the following video (its a technique I think some are calling “direct modelling”):

Also, if jameg2169 (the originator of the quote) happens to stop by, the part about traditional modellers being good at subtle surface changes… to me, that’s exactly what parametric modelling is great at. However, radical changes to surfaces usually means you have to rebuild huge parts of your model if you go with parametric, wheras using, for example, a subdivision surface modeller, you just push and pull all you want (though I realize that that particular technique isn’t used for ID, and absolutely not cars).

Well, it all depends on what you use it for. I still value the freedom 3ds Max gives me, those kinds of packages will get you geometry. on screen. fast! When it comes to detailing, things get more complicated. That sounds like a perfectly logical trade-off for film and games (which are its primary markets)

For design I agree some freedom must be sacrificed for quality, however, parametric modelers take this to the extreme. You can’t edit anything on the fly (pretty much anyway) and unless you got lucky that a certain dimension or feature has the value you just happen to want to change, you’ll need to put more effort into making your design change than you perhaps should.

Now don’t get me wrong, I’m loving parametrics more and more myself. At first it was like “first a plane…then a sketch… then constraints…and then finally it’s 3D” but now I’m starting to think more like “ok, let’s make a plane to draw that 2D profile on there and just loft it through”. It’s easier, it’s not tedious extra steps, its appeal is definitely warranted.

But to say parametrics modelling is the wave of the future? I don’t think so, personally.
Pro/Engineer was one of the first popular parametric applications, and it came out in like 1987.
If parametrics were the wave of the future for all modeling, other modeling methods would have gone by now :slight_smile:

Theres no such thing as “the best” tool. What there is is a set of existing tools that are optimized for particular situations.

Pro E and other tools like Solidworks have come a long way and with plugins like ISDX offer a LOT of functionality and flexibility.

In the real world you’ll generally find there is rarely a “1 stop shop” for CAD.

I use Alias because it offers me an extreme amount of power and flexibility in my surfacing that even using IDSX I have not been able to come close to how efficient I have been.

I’ve seen manufacturers try to “recreate” my designs in Pro E and even after 4 or 5 revisions (which usually take several days) they still can only come close, because they don’t have the ability to influence transitions or complex curvature nearly as easily as I can in Alias.

I also use Pro E. Once I finish my surfacing in Alias I import them into Pro and can then do all of my “engineering” features like offsets, reveals, and create a clean solid model that retains my design intent and is now completely worthy of going straight to tooling. No rebuilding, no cleanup, just class A exterior surfaces with the flexibility of parametrics. If I find out later I want to change a transition I can export a new surface, replace my existing patch, and if I play my cards right all of my offsets and features will rebuild based on my new geometry.

It’s ridiculously complex - but it works and works VERY well.

If parametric modeling is the wave of the future, then that ship has sailed. :smiley: All fun and hyperbole aside, Cyberdemon hit it square on. in my 12+ years of working in the design world I have yet to see a single tool that can do it all. I completely agree too that parametric can’t do the type of subtle detail in surface definition that a nurbs based tool, like Autodesk Alias Design, can. Y simply cannot easily do the type of surface manipulation in something like swx or pro that you can in Alias. That being said, however, at the end of the day it’s about getting your work done and the tool that helps you to do that most effectively (or the one that is your companies standard) is the one you’re going to use.

If you want to see the future of parametric modeling look at the direct modeling stuff being done with Autodesk Inventor Fusion, Siemens Synchronous Technology, Sketchup. They work on slightly different paradigms, but this is really what you will be seeing a lot more of in the CAD world. I would argue though that this will not be a full blown replacement of parametric. After all, how many people out in the world are still using 2D CAD?

Great discussion.

BTW @Cyberdemon you should see some of the Digital Prototyping solutions we have at Autodesk especially our Alias to Inventor interoperability. Open your Alias wire files natively in your MCAD tool. . .Dare to dream.

best.

j.-

I think there are a lot of cool technologies coming out.

I HATE Pro E. With as much passion as one could. But as long as every Chinese vendor is running a pirated copy of it, it makes switching hard.

Very true. Lots, I’ll bet.

And it has a good reason: 3D software is very specialized and marketing demos always make it looks easy but once a company actually gets it in their hands, I’ll bet just shy of a quarter keeps it for enough time to become apt at it.

I think people come up with a workflow that works for them and they go with it. Who would think a designer could prove form better using a parametric modeler over a non parametric modeler like Alias or Rhino? That just means that the designer has not discovered that workflow yet.

And just because the tool has the words ‘engineer’ in it does not mean that it is just a tool for engineers. Like Catia Pro/ENGINEER has a set of industrial design tools. PTC Sales people sell it as ISDX. It is not that simple however… The real power of Pro/ENGINEER… call it wildfire because ID’ers can get hung up on that word ‘engineer’ so from now on we will call it ‘Wildfire’.

The real power of Wildfire is proving form: the combination of Parametric curves and surfaces that work with parent child relationships working together with ISDX curves and surfaces. be careful having your PTC reseller show you this functionality because many are not prepared nor do they know what ID is.


And as for as the word ‘hate’. That is a powerful word and should be used only sparingly. Hate is for close minded individuals.

“That just means that the designer has not discovered that workflow yet.”

Or that this workflow doesn’t suits his ?

I don’t want to offend you Bart, from your posts you seems to be the godfather of ProE, I would love to have such a nice teacher (for other softwares) but as you are an expert in ProE you are obviously biased.

I personally believe [doesn’t this sound a little bit like Miss South Carolina ? : ) ] that the designer, when looking for a form or a design or a concept, needs to be free from linear analytical thinking "if a build this like that then like that I’ll be able to modify this using that later on…) No. A designer prefers the crude “no history whatsoever” Rhino because it’s simpler and if he as to redo it well its a pain in the ass but at least the operations are simpler enough for him to care 90% about his design and 10% about the software. Actually building the surfaces isn’t what takes the most time. Its finding the right surfaces the long path. Redoing something you just did, either because of a crash, or a change of mind in a non parametric software, just takes minutes. Finding the right design takes hours. Everybody (every IDer) has experienced that.

Now we can discuss this. I’m sure you will : )

I just can’t imagine why an IDer who is paid to do the visble parts of an object/product would use an overly complicated program, like the big three (worldwide) ProE, NX, Catia, intended to do the intricate inner parts. If your job is to do the inside of a turbine at Rolls Royce Aerospace you are not an IDer but an engineer. Close but not same.

Sure you personally can make everything with ProE. But you have been practicising full time for ten years ! I don’t have that much time.

Georgeous designs have been created using pen and paper. Let’s say the Jaguar type E. Or the less known type 13. And now using the amazing Wildfire there are countless ugly boxes. Talent always overrules tools.

Oh wait thats not true. No one car design was made using Wildfire. They use Alias. No construction history to speak about. Poor format translation. Etc… Curious. Does that means thousand of people around the world “haven’t discoverend that workflow yet” ?

Call me daft, but reading Sleek’s post just made me realize why people use clay.

Bart taught me an incredible amount of Wildfire in three weeks. It doesn’t take ten years. I’ve been using it for ID every day for five years now. If you do not know any Pro/E, or very little Pro/E, you should stop criticizing it because you have absolutely no clue how much power that software has. If you harness this power and use it creatively, the sky is the limit and an awful lot of time can be saved in the design to manufacturing process.

Learn as much as you can about as much as you can, man.

Ask an Illustrator user the last time they hit Ctrl+Z… chances are it will be seconds ago. And the right design may not take hours, it may take minutes, or days or years. You never know. And limiting yourself to one method for exploration, surface creation, etc. might limit your design. Might…

Parametric modeling might be the wave of the future of CAD for final production. SolidWorks, for example, certainly isn’t to the level of surface manipulation and subtlety of Alias yet, but it is getting better. For most, not all, products it’s probably already to a level where the consumer wouldn’t notice if you recreated an Alias model in SW.

But even if parametric modelers get astoundingly better at surface manipulation and subtlety I still don’t see products like Alias or Rhino going away (and I’m a SW guy). I think I’m pretty good at SolidWorks, but on my latest project the surfaces were somewhat complex, and so much thought was being put into constructing the CAD model that my attention was drawn away from properly filling in the 3D aesthetic details that weren’t in the 2D sketches I’d made previously. So when I was “done” I had to take a hard look at the model, then go back and tweak everything (even though aesthetics were certainly not at the top of the client’s “must have” list.) And start to finish that took awhile. So while SW might’ve been the right tool for this job overall (if for nothing but making sure my surface were always drafted with the model’s weird pull direction), I think it does point out that SW, and probably parametric modelers, might not be the best for: A)quick models that aren’t a cube, B)exploring 3D geometry or C) creating complex, subtle, or freeform surfaces. I hope someday I’m wrong about C, because I love being able to go back and edit over and over again the sketch that helps define some curvy surface- even edit it to be what it was originally!

I don’t mean to bash SW too much, after all it is my preferred software. I think it’s great for a lot, if not most, things. And as an anecdote, in college I remember actually being more frustrated with Alias than SW, because everything seemed so loose and hard to pin a dimension on, which to me doesn’t speak to final geometry modeling. Perhaps the future really lies in the integration of both techniques (as is hinted above), either by better sharing of data between separate programs (Alias <–> Inventor) or having one be a sub-program of the other (ISDX style).

“If you do not know any Pro/E, or very little Pro/E, you should stop criticizing it because you have absolutely no clue how much power that software has. If you harness this power and use it creatively, the sky is the limit and an awful lot of time can be saved in the design to manufacturing process”

Yes, true. I don’t use ProE. I’m saying this based on seeing R&D people using NX and ProE and SolidEdge (poor software this one imho). Seems complicated to me. When I ask them (face to face)for a modification I don’t have that “let’s do it attitude” I have with Solidworks folks. I rather get an annoyed look and a “Maybe I’ll do it as you say and have a screenshot/step/iges send to you ? Later on ?” Meaning, to me, “Oh my god I just can’t touch this cards castle in real time. I will try to do it but I sure can’t do it live.”
But maybe those guys using SWX are good and those using NX…not as good ?

But still great discussion !

But maybe those guys using SWX are good and those using NX…not as good ?

I work with both Alias and SW masters and they all can rock out some (complicated) 3d geometry fast.

anyone who builds a parametric model and then is afraid to go back into the tree and change something…isn’t a very good CAD Jockey. The whole idea of parametric is that you can use that history tree to manipulate the geometry.

There are limits of course…but if built and referenced correctly…one should be able to tweak until they are blue in the face.

all tools have pros and cons and each one seems complicated (because they all are!).
I’ve used Rhino, SW, pro/e, UGS NX and even 3ds max to build geometry in a production environment. I’ve been trying to get more face time with Alias because it interests me and I can easily say the Alias has been the hardest to learn. There seems to be a tool for everything…which is amazing and I can see why so many people love it for its speed…once you know where everything is.

Rhino + Grasshopper = parametric heaven for me. I find that I can model and describe conditional relationships very intuitively, quickly, and with high precision. And yes, parametric modeling is the wave of the future, if only for the fact that it allows for a more inuitive and immediate response to a large variety of meaningful information while modeling.

Question: isn’t a STEP or IGS file the same no matter what program it comes from? If so, and I believe they are, then it’s really irrelevant if it comes from a parametric program or not. I think a lot of people on here like to state that this or that software is supreme in order to lift themselves up. There are a lot of 3D design programs out there…competition.

That said, I’ve used both. Rhino vs. Solidworks. Rhino is FANTASTIC at conceptualizing ideas on screen. I’m actually using it right now to see how shapes interact with each other following a thorough sketch-o-rama. By doing this, I can see how things truely interact and can revise my sketches accordingly. Depending on the complexity, I’ll either go with Rhino or Solidworks at this point to develop the actual CAD model. If it’s a complex surface, I might create it in Rhino and export over to Solidworks. If it’s only a skin, I’ll stay in Rhino. If it’s injection-molded and needs inner walls and multiple, complex fillets, I’ll send it on over to Solidworks. Just depends on what I want to do.

Im actually surprised that Rhino and Alias have not incorporated parametric controls. I remember back when Catia did not use parametric and they lost customer by the hundreds each month. Now they have a hybrid set of tools that leverage both non and parametric tools. Rhino is still struggling to get parent child relationships to work still after so many years… WTF

How is the new year treating you guys looking for jobs? Is the new year picking up inquiries for interviews?

The question is hard to answer. Regardless of the format, the real question “Is a surface always a surface?” Yes and no.

You can have two surfaces that LOOK identical, but one may be significantly more complex than another. This is one thing I face daily when working with geometry that gets created and pulled out of Pro E, vs data I create in Alias. In my experience Rhino is also very good at creating a surface from just about anything, but often times those surfaces are also more complex then they need to be. A lot of this has to do with modelling practices, but ultimately the ideal goal is to have your geometry be as SIMPLE as it possibly can be. This isn’t always realistic, necessary, or achievable, but the simpler and cleaner your geometry is the easier it will be down stream to edit, modify, use for calculations, etc.

So the ultimate answer to your question is “It depends” A step file is always a step file, yes. But anyone who spends a lot of time doing data exchange can tell you sometimes that geometry can be MUCH uglier and difficult to work with depending on how it was created and in what application. Rhino in my experience does seem to be the best at exchanging data, but from a surfacing standpoint it has a lot of tools that will create the surface you want, but with much more complex geometry than is needed.

I actually just came across Grasshopper the other day, and was definitely intrigued. However, my understanding (from a brief look at their website only), was that it was for “exploring new shapes using generative algorithms” - basically creating patterns and equation driven models that would be difficult to do otherwise, not creating the “basic” forms that parametric modelers usually make. And not for managing the entire model, just the special algorithm based surfaces. I was interested when I thought it was just the fun generative algorithms (though I doubt I’d get to play with that in my current job :-/ ), but now I’m even more curious. What do you use it for? Funky shapes or boxy toasters?

Grasshopper is the incorporation of parametric controls into Rhino.

Here is the website for users, which includes a lot of interesting stuff. David Rutten, who writes the software, is always on the forum answering questions and getting feedback on the sotware. Most of the work you see will be for funky shapes, but I know of people who use it to create full parametric models for large building projects.
I use it constantly from everything from just adding functionality to Rhino (like piping a bunch of lines at once) to this project. The interface for Grasshopper is amazing. It makes other software feel old and awkward. And it’s FREE. The latest work in progress allows you to edit the logic structure of your parametric models in a really easy way, using a tool called PathMapper. You can read excel data into it, or feed data to excel sheets. Andrew Payne has used it to make G Code for CNC milling. You can make really simple parametric models in a way that is more intuitive than any other parametric software I know of, but you can also extend it with VB.NET or C# to become a powerhouse of information-based modeling. I can’t recommend it enough.