Issue #1,107 | Inside the Business of CAD | 4 October 2021
You’ve probably seen CAD demos that start with a square box covered with grid lines, which are then prodded and pulled to arrive at a curvaceous shape. This is sub-d (sub-division) modeling.
In the figure above, the vertical and horizontal lines on the body lay out the subdivisions. The lines in blue are being interactively pushed (by 5.24mm), much like in clay making. The body is a surface model, which can be converted to a solid, if not too complex.
We are so used to working with the precision afforded by profiles and solids (b-rep modeling) that the slapdash nature of subdivisions seems like an unhelpful distraction. Nevertheless, it has its place in the designer’s toolbox.
Here Jonas Kunze looks at the pros and cons of using sub-d modeling for our design work.
- - -
Brilliant concepts need brilliant tools. Otherwise, creative and innovative ideas can fizzle out soon after they emerge. This raises an important question: do designers today have tools that match the creative flair inherent in design as a discipline? I examine the question with reference to sub-division (sub-d) modeling.
What Is Sub-D Modeling?
Subdivision surface modeling is a concept with which designers may well be familiar, since this technology represents the state-of-the-art in digital modeling. The technique comes from character design, and was originally introduced by animation studio Pixar as a faster way to model animated characters and CGI [computer-generated imagery].
With sub-d modeling, shapes are achieved by manipulating control points through several kinds of push and pull actions. The technology is incorporated into a number of software tools used in design, such as freeware Blender (see figure below) and Dassault X-Shape (see figure above).
Advantages to Sub-D Modeling
Sub-d quickly gained popularity due to the time-saving it offers. Its use then expanded beyond character animation (such as in computer-animated cartoons and games) to other industries, such as industrial and automotive design.
The technique includes the following advantages:
-
Time efficiency. Subdivision modeling is a quick approach to sketching. Once users are familiar with the software, results can be achieved with substantial time savings, especially when compared to other tools like NURBS [non-uniform rational B-splines].
(For more on this subject, see “Surface Modeling Techniques For Automotive and Product Design” at peer.asee.org/surface-modeling-techniques-for-automotive-and-product-design.pdf.)
-
Uniformity. Sub-d performs well when designing organic shapes (such as cartoon dinosaurs) and where little variation is required; this is why it was created originally for animated cartoons. This makes it a fitting tool where overarching symmetry and uniform design results must be created fast.
However, results may be less than ideal when wanting to imprint a unique identity on a product.
-
Convenience and familiarity. The intuitive nature of sketching in sub-d makes it a convenient choice for designers who are familiar with NURBS tools in CAD. Sub-d is similar to polygonal modeling, making the familiarity factor a plus. For many designers, sub-d is the go-to solution whenever objects cannot be easily modeled with NURBS in CAD, for example in the case of styling surfaces (see figure below).
(For more on this, see rhino3d.co.uk/rhino-for-windows/an-introduction-to-subd-subdivision-surface-modelling-in-rhino3d-v7.)
Overall, we can say that subdivision modeling is suited to tasks where quick visualization results are more important than precision.
Disadvantages of Sub-D
We cannot forget that this modeling system was originally developed to recreate organic shapes (a.k.a. human and other non-prismatic shapes). This is very different from the kinds of inanimate objects common to mechanical design, explaining sub-d’s short-comings:
Fast to Slow. Speed is not a constant in subdivision modeling. While getting an initial model takes little time, there is often a speed mismatch between the subsequent stages of the design process; progress in later stages slows down. Sub-d eventually falls short in automotive design, where renditions of concepts become increasingly detailed as ideas are discussed, worked on, and reworked.
Lack of Precision. Speed is important in commercially-oriented processes, but there is a price to pay for sub-d’s time savings: lack of precision. Control and precision are paramount in automotive and other designs as they play a crucial role in defining value and impact the product’s final price. When it comes to using sub-d, there is an implicit assumption that designers will have to accommodate a certain degree of inaccuracy.
Non-iterative. Subdivision modeling does not lend itself well to the iteration loops needed in the automotive design process, because every change requires starting from scratch. As others have stated, one of the shortcomings of this technique is the lack of continuity. So, to add details or make changes in later stages, designers resort to other tools. In this sense, sub-d inserts disconnection and fragmentation into the process.
(For more on sub-d modeling, see Matt Lombard’s article “What Is SubD Anyway?” at dezignstuff.com/what-is-subd-anyway.)
Does Sub-D Have a Future in 3D Modeling?
Many designers agree that modeling tools currently available have important limitations. In the case of sub-d, we cannot deny its usefulness, especially for first-stage visualization purposes. But given its shortcomings, can this technology be enriched, complemented, or replaced? One area where sub-d works well is in interactive virtual reality modeling.
In the absence of a better option, sub-d is the fastest way to generate the impression of an idea. It is certainly handy for visualizing concepts to assess their feasibility. But since this technology is admittedly lacking in dimensional accuracy, is it possible that it has outlived its usefulness?
[Jonas Kunze is a co-founder of Flying Shapes. This article was first published on https://www.flyingshapes.com/designing-with-subd and is reprinted by permission, with edits.]
== 3D Data Translation Software Tailored to Solve Your Problems ==
Education and awareness make for flawless, error-free, and numerically-correct MCAD conversions!
Okino’s PolyTrans|CAD software solves complex, demanding 3D conversion problems by transforming CAD and DCC datasets into highly optimized and efficient files for all common downstream 3D programs and authoring packages. Our staff ensures that each Okino solution fits your conversion problems 'like a glove', including useful discussions on how to source the best 3D datasets.
Popular CAD data sources we support include SolidWorks, ProE/Creo, Inventor, AutoCAD, Revit, Navisworks, DGN, IGES, STEP, Parasolid, and JT. DCC data sources are Cinema-4D, 3ds Max, Maya, FBX/Collada, and many more.
Perfected over three decades, we know 3D data translation intimately, providing you with highly personalized solutions, education, and communication. Contact CTO Robert Lansdale at [email protected].
And in Other News
CADENAS’ 3DfindIT.com search engine integrates 3D CAD models directly into video meeting platforms like Zoom and GoToMeeting.
Models from 3DfindIT.com are inserted into VidiPlus via drag’n drop, which then appear on the screen for everyone to see. CAD models can be sliced, rotated, zoomed, and made transparent. More at www.cadenas.de and vidiplus.com.
- - -
Siemens last week expanded its Xcelerator offering by launching “Xcelerator as a Service” or XaaS, software based on MindSphere, Mendix, Supplyframe, and TeamCenter X. It offers collaboration, mechanical design, electronics, and software development on the cloud. sw.siemens.com/en-US/digital-transformation/cloud
- - -
QuadriSpace releases Document3D Suite 2022 consisting of Pages3D for technical documentation and Publisher3D for technical illustrations. The new Detail View Markup creates illustrations with zoomed views of 3D models.
A new part filter sorts and identifies parts meeting user-defined criteria. Part metadata can be imported from and exported to Excel for access to ERP and PLM data. Thirty-day demo (after registration) and more info from quadrispace.com/products/whats_new_2022.
Letters to the Editor
I’ve been watching the consolidation of the CAM [computer-aided manufacturing] industry with great interest.
Companies have acquired an extensive portfolio of CAM products, but for the most part have been unable to consolidate them, due to their inherent proprietary nature. The proprietary approach ensures that manufacturing knowledge encoded into each resides solely in each.
Much like CAD systems of the past were faced with a similar proprietary representation which locked users into their platform, which encouraged [intermediary] data exchange formats, this CAM consolidation could also be the motivation for a CAM interoperability standard. Such a standard does exist in STEP-NC, an ISO standard which would provide a neutral file format to exchange process data between CAM systems.
Maybe the time is right for STEP-NC -- depending on whether the acquirers are looking for ways to consolidate their CAM assets.
- John Callen
The editor replies: Software companies that are acquiring CAM software would want something like STEP-NC, but I am finding that data integration between acquired companies (under the larger entity) is an inexplicably slow process. (Nemetschek Group might be the poster child for this.) Perhaps this is why Autodesk rips its acquired software apart!
Investment companies that acquire CAM software would not care about integration.
Mr Callen responds: You're right. Investment companies that just want to milk the cash cow, and would not care about interoperability within their portfolio.
Though STEP-NC could provide interoperability across a vendor’s CAM portfolio, it would also potentially allow migration to a CAM product outside the portfolio. MCAD went through a similar cannibalization scare, which really didn't materialize. Users still like to use what they like to use.
Re: A.I.’s Fatal Flaw
In upFront.eZine issue 1,105, you wrote “The blackbox crutch executives can say decisions were made by AI, and so deny culpability.” This is the Sergeant Schultz Defense (“I know nussink”) and it usually fails. See en.wikipedia.org/wiki/Idiot_defense
- Don Beaton
- - -
For all my techie friends: a great article by CAD’s equivalent of Gandalf the Grey. What is AI, and what has *really* been accomplished so far?
- MattyKUSA via Twitter
- - -
You wrote: “We see obsessiveness over AI because it is the ultimate computing problem.” This is truly a factor, but so are economics. Investments in technology are made in hopes of strong returns on capital. Technology has profitably replaced a large chunk of the muscular tasks that were once the province of humans.
The big economic opportunities now lie in replacing intellectual labor, which is generally expensive labor. Computers replaced plenty of intellectual work already. I once worked in an insurance company; by 1980, the sea of clerks of the earlier era had largely been replaced by mainframes and a lake of programmers. Now the less repetitive, more complex intellectual labor is where the opportunities are.
As for CAD, while I am sure that AI will power certain tasks, AI is far from being able to do the core work, which is design. Certainly in AEC, we are far from organizing all the disparate knowledge that underlies a design in a manner such that AI can use it.
We already see this problem in BIM, where BIM evangelists and salespeople want to credit BIM for benefits that substantially arise from Integrated Project Delivery (IPD). IPD is the only current way we have to get all the needed knowledge into the design, which is not to ignore things like automated checks for building code compliance.
Lastly, as a species, we are going through plenty of changes of what it means to be human. A few years ago I was astonished to read that significant numbers of teens were more comfortable texting their friends than speaking with them face-to-face.
Back in 1976, Joseph Weizenbaum published Computer Power and Human Reason. In the mid 1960s he had invented (designed) what we today call a chatbot. ELIZA was a very early piece of AI (See en.wikipedia.org/wiki/ELIZA for details.)
At that time it operated using a teletype machine interface. In one version ELIZA acted like a psychotherapist. Weizenbaum was struck, and alarmed, when his secretary developed a strong emotional attachment to the program. Thus his book, which raises the now-old and still unresolved question of how we will be changed by what we have created.
- Leo Schlosberg
The editor replies: I remember running Eliza on the university mainframe computer at in the late 1970s. It very quickly became repetitive in its answers.
I don’t think that humans change in reaction to new technology; we adapt. For instance, our fingers gained muscle memory for touch typing, and then for typing on small glass screens.
Mr Schlosberg responds: What got Weizenbaum going was the secretary’s (and other people as well) reaction and attachment.
The specific version was one that mimicked Rogerian therapy (pioneered by Carl Rogers, also known as “nondirective therapy” or “client-centered therapy,” and based on pretty much rephrasing what the person said in an affirming way or asking modest questions, like “How did that make you feel?”). That may be the ELIZA you met but possibly it was a different version.
I almost wrote you today after seeing a few articles on AI. One was in an archaeology magazine about an application in which the AI was better than archaeologists at matching pottery shards. That, plus your article got me to checking in on radiologists, who surely are ripe for substantial displacement/replacement by AI.It was in a professional publication and not very incisive.
The editor replies: Humans yearn for non-human attachments, because they are faithful and cause no problems, the way that other humans do — like a dog, but without the problems dogs create.
- - -
Nobody is better than Ralph Grabowski at telling the emperor of software he has no clothes. Worried your job will be replaced by a computer? Take 5 minutes and read Ralph Grabowski’s "AI's Fatal Flaw" and you’ll feel a lot smarter.
- Roopinder Tara via Twitter
- - -
With Minsky, perhaps we just misinterpreted his “Within a generation.” Assuming the human race survives, it will happen within a generation. We just have no idea which one! :)
- Robin Capper
- - -
I've been relying on Ralph Grabowski's commentary on the CAD industry for nearly three decades just myself. He never fails to provide something new. Here, I learned two new terms that describe the AI crutch:
-
The Black Box Crutch
-
The Boombox Crutch
- Darren Young via Linked-In
- - -
Is today’s AI just a collection of rules and parameters, that the computer uses as boundaries and never forgets? If so, should we not call it artificial ‘intelligence’, but just ‘artificial rote memorization’, or “ARM” ?
- Peter Lawton
The editor replies: True AI would make inferences from data presented. Problem is, if it were to do that, we would have no idea how it arrived at its conclusion, and so would be unable to confirm if it was correct.
- - -
Another eloquent and insightful piece of writing. Common sense in this industry is getting to be uncommon.
- Jess Davis
- - -
I agree with most of what you say about AI. I do believe that advances are being made. I saw this on a blog and thought it was pretty amazing: “OpenAI Codex Live Demo.”
- Conan Witzel
- - -
A friend of mine said around 1990 the ultimate sentence about AI, which has been true since then, and always will be, more so than Moor´e’s Law: “Artificial Intelligence is everything computers still cannot do.”
- Fernando Valderrama
- - -
I have enjoyed reading your upFront.eZine for most of its 21 years. Having corresponded with you previously, I feel you will find the discussion linked below to be interesting and informative: Online Conversation: Understanding Transhumanism.
Best line in your article: “What is AI’s fatal flaw? It cannot replicate human metaphysics; the rational cannot conceive of the irrational.” Many thanks for your excellent journalistic efforts.
- Harold Chaney
The editor replies: The video notes that an alternative name proposed for AI had been “complex information processing,” which would be more accurate, but that “artificial intelligence” was considered more aspirational. “There is something about a term about something that we can’t achieve that inspires people to think beyond existing constraints.”
- - -
Your writing about AI in your latest issue is excellent. Thank you for your voice of human rationality (or is it human irrationality, or both?). A much needed dose of whatever it is (intelligence).
- Rob Snyder
Notable Quotable
“If people yelling on Twitter can cause policy change at Apple, imagine what will happen when nation states exert pressure to increase the scope of on-device searches.”
- Diogo Mónica
Thank You, Readers
Thank you to readers who donate towards the operation of upFront.eZine:
-
Michael Kuzmik
-
Jeremy Powell at Vectorworks (large company donation): “Keep up the great work on upFront.eZine, Ralph!”
To support upFront.eZine through PayPal.me, then I suggest the following amounts:
-
$25 for individuals > paypal.me/upfrontezine/25
-
$150 for small companies > paypal.me/upfrontezine/150
-
$750 for large companies > paypal.me/upfrontezine/750
Should Paypal.me not operate in your country, then please use www.paypal.com and use the account of [email protected].
Or ask [email protected] about making a direct bank transfer through Wise (Transferwise).
Or mail a cheque (US$ or CDN$ only, please) to upFront.eZine Publishing, Ltd., 34486 Donlyn Avenue, Abbotsford BC, V2S 4W7, Canada.
*4710
Comments