Organic Modeling with Engineering Precision
Autodesk left SolidWorks reeling after purchasing Tsplines. Tsplines, for those who don’t remember is a technology that lies between mesh modeling and NURBS modeling. It allows the worlds of 3DS Max and SolidWorks to merge a little bit. I believe that this is the direction of the future of CAD. More than “3D EXPERIENCE”, more than “the Cloud”, more than mobile, or SAAS , or cross-platform, all of that is either nonsense or IT infrastructure. It’s debatable whether any of that could be called innovative, but it’s surely not related to CAD in any way. There are two things going on in CAD innovation today that are related to CAD, and one of them is merging mesh techniques and NURBS accuracy. A quick demo of what I’m talking about with “mesh” techniques is here. It’s also called subdivision modeling, subd, and there are a lot of different tools that do this, like Maya, 3dsMax, Mudbox, Zbrush, Sculptris, modo, and more. This stuff is where CGI comes from, animated movies, computer games, 3D character animation, and a whole world of “modeling” that is 3D but typically not associated with CAD.
Of course mesh techniques are really just where direct edit intersects organic shapes. So the second thing going on in CAD today is direct edit. History modeling has some advantages, and to be clear, I advocate a combination of history and direct, like what Siemens has done with Synchronous Technology. The reason I see direct edit and mesh modeling as related seems obvious to me, but maybe not to others. They both relate to directly interfacing with the geometry, and using the computer interface to change it. They don’t make some abstract level between the interface and the geometry, like a system of sketches, features, and settings. In both direct edit and mesh model, the software has the smarts, and the data is dumb. In history modeling, the data is “smart”.
There are different levels of sophistication of these tools, and the link above to a Sculptris demo shows what I think is fairly sophisticated. Less sophisticated tools (from the mesh modeling side) are things like Tsplines and this Power Surfacing from IntegrityWare. It’s all about selecting controls, and moving the controls. Notice that Sculptris does away with the controls and the selection methods, using “brushes” with various properties.
)/newsletter/2009_august/mg.jpg?resize=463%2C258)
There are several reasons why I think these two worlds (subdivision/mesh and NURBS) need to collide. One is that the direct edit tools out there prove that a more direct interface works. You can still have parametric models, models driven by numbers, even if the interface is more direct, and less abstract. And second, CAD needs more than just prismatic shapes. Subdivision modeling shows that organic shapes are easy to create, while they are notoriously difficult in history-based CAD, and analysis tools show us that organic shapes are in many cases more efficient than prismatic shapes for stress distribution, aerodynamics, ergonomics, shape optimization, and other analytical factors. This is why I believe engineering will involve more and more organic shapes in the future, and that the current state of CAD is really what is holding us back. I think CAD needs to rise to the challenge of the geometrical future, and stop hiding behind IT silliness of hardware, networks, cloud, OSes, mobile devices, and what not. I’m hoping some of these new CAD development teams like Belmont, or any of the Russian teams working toward more options in CAD can re-engage in CAD, re-imagine CAD, and give us the computational power we need to move forward.
The CAD software SolidThinking has a module called Inspired that will take your constraints and build an optimized shape around it. Optimization rarely forms prismatic shapes. It’s usually organic. Bones do not grow in straight and cylindrical sections.
Mesh modeling is also related to things like reverse engineering, laser scanning, 3D printing, medical data from MRI and CAT scans. So its a wonder to me that this form of data has been ignored so long, especially when it has so much inherent value for CAD users and mechanical engineering applications. Especially in an age when spending on biomechanical engineering is so high.
I get accused of being afraid of change from time to time. And maybe I am. I’m afraid of change that doesn’t really accomplish anything, or where the change is mostly irrelevant. Most CAD products today are built around the machine design industry, and this focuses on easy to manufacture shapes, which are typically prismatic. I know accomplishing anything in industry today is all about money, I wish it were more about progress, skills, capability, quality of life. Machine design is where most of the money is in CAD. Developing software that engages too much of a niche is seen as risky, but if well executed would pay off. This is why it has taken so long to get to the point where mesh modeling is starting to be taken seriously in CAD.
Tools like Tsplines have the seeds of what we are going to need to engineer great stuff in the future, but just the seeds. Tsplines is not really a tool yet. Tsplines generated a lot of attention for SolidWorks, and Autodesk yanking that rug out from under them shouldn’t have been unexpected. Moldflow was a huge industry player for plastic part manufacturing analysis, and the same thing happened there. Linius, back long before these, was a wire harness partner software that Autodesk also purchased, and even tools like Alias, Maya, and many others. So this kind of thing has been going on a long time, and for SolidWorks to get surprised again and again points out what should be obvious. SolidWorks reaction to the Tsplines sale is this Power Surfacing from Integrity Ware. “What?” and “Who?” I hear you asking. SolidWorks had to do something with all of that egg on their face, and Power Surfacing is at least attempting to fill the role that Tsplines left empty. Not sure if DS participated at all in the development, with personnel or financially.
There is an old interview over on the SolidSmack site where Bernard Charles says that Catia has “the best surfacing in the world… By Far” he says with a knowing grin. But that’s not doing any of us SolidWorks users any good, except for the Fill surface. It could be that part of the plan is to make Catia surfacing available to SolidWorks users, but who knows at this point. Even the Power Surfacing software isn’t enough. The capability has to be native. Power Surfacing has no ability to edit SolidWorks surfaces, so there is no interoperability. You’ve got the ability to make blobs, and the ability to make specific shapes driven by numbers and splines. But you can’t combine the two.
This is the innovation I’m looking for from the CAD industry. At this point I don’t care who. I just want real organic modeling that can be controlled with engineering precision.
That’s great! Amazing. You can also check out this video https://www.youtube.com/watch?v=S3L_W0KW18I&list=PLXHTPs03CIrhWHKpKBpncxsA942_VaWlh&index=18 on how to use SelfCAD to make unique organic 3D shapes. SelfCAD is a great software, suitable for both beginners and experts. Check it out.
Matt
I am curious about subdivision modeling, like why anyone doing product design wants it. I have not seen any real products that I know were designed with subdivision modeling. It looks like a great tool if you are designing monsters, but consumer products not so much, unless there’s a new aesthetic I am missing. That bike in your article has no appeal for me.
I tried a demo of Solidworks Industrial Designer a few years back, at a World event, and I thought using it was a wretched experience. There’s a lot talk about subdivision modeling, so I figure it must be meaningful to someone. It would really help if I could see some examples of actual injection modeled consumer products that were designed that way. That would give me some incentive to try ti again.
I am an Industrial Designer, RISD 1978 (yes, way back), and I have been an in-house designer since then. I’d like to keep up with things, if I think it might pay off.
Thanks
Dwight
Dwight,
I’d have a look at a couple of things: Power Surfacing https://www.npowersoftware.com/NewPowerSurfacingOverview.html and Tsplines (essentially Fusion 360) https://www.autodesk.com/campaigns/fusion-360-design?mktvar002=1028432&mkwid=s6lVBtlie|pcrid|308334740598|pkw|tsplines|pmt|e|pdv|c|slid||pgrid|64140220787|ptaid|kwd-371120189250|&intent=&utm_medium=cpc&utm_source=google&utm_campaign=GGL_Fusion+360_US_NB_SEM_EXACT&utm_term=tsplines&utm_content=s6lVBtlie|pcrid|308334740598|pkw|tsplines|pmt|e|pdv|c|slid||pgrid|64140220787|ptaid|kwd-371120189250|&addisttype=g&s_kwcid=AL!8131199977!3!308334740598!e!!g!!tsplines&gclid=CjwKCAiAyfvhBRBsEiwAe2t_i9NP9FlREX9gyRRNcm-UF1Omy6wCZut2k3cOq9i_7-xNKl_V0IXlzRoCCEcQAvD_BwE
The non-history, non-parametric, non-feature based mesh type modeling generally makes the outer shape, and then you might use something else for the shell and engineering work. But there has to be a way to go between mesh and NURBS.
ViaCAD Pro and SharkCAD are retail and mid-range CAD products Mac and PC, and the PC version particularly of Shark has some great value features like SW/SE/CATIA4_5_6/ProE/NX/Parasolid exchange. These are DS translators that aren’t available for Mac, apparently. But the PowerPackPro plugin uses the IntegrityWare/NPower libraries to do mesh to analytic surfaces (planar/cylindrical/conical/spherical sections) and mesh to nurb objects. It can convert a manifold mesh object into a manifold curvature-continuous nurb object. VC and Shark have subD modelling inbuilt, and the nurb equivalent object can be kept associative to the subD, so it updates if the mesh does. It’s a live interface between CAD object and freeform subD surfaces, It’s not possible to import meshes and have them work that way, the IntegrityWare software like predominantly quad meshes, so triangular formats like .stl have to be retopologised, a quad mesh is fitted over the underlying triangle mesh. I’ve notice that the IntegrityWare plugin for OnShape has that, there are others in subD modellers like 3DCoat (automatic) and Blender, manual. I do think that Shark and VC are a step forwards towards subD and CAD harmonisation and integration.
A couple years ago I did some milling on a project using the ATOS system. We used an ATOS III to scan a bulkhead and machine another bulkhead covered in phenolic honeycomb to get a perfect, void free mate. This bulkhead was a 20′ diameter dome. Our tolerance goal was about .002″, ended up being able to scan about a square meter at a time (coverage per scan based on res). We had to also use their Tritop software to stitch the pieces, output to Catia. Worked great, and I don’t think there is a anything else out there that can do something like this, this easily. Scanning bright, shiny aluminum is tough. We had to use a few targets per square foot to aid in the alignment. Cost about 250k for the entire package, though. I don’t have any links to the scanning, but the machining is done here, 1:20 into the video.
http://www.youtube.com/watch?v=zWttk8eMxNo&feature=channel
Off topic, but have you seen the GOM Atos optical scanners?
http://www.gom.com/
I was lucky enough to see a demo at a clients studio (they purchased one, lucky them!)
It was the most pain free scanning I have seen yet, overlayed and matched each survey over the last to make a watertight scan.
Of course, not cheap…
@matt
Yeah, the reason SensAble (now Geomagic) works so well on scanned data is because it is NOT polygon based. It is based on Voxels (spatial occupancy grid — kinda like a 3D pixel — where material is either there or it isn’t).
Funny you mentioned a baby doll — one of our first projects when working on the NURBS wrapping technology at SensAble was a baby doll head from Hasbro. Their guy sculpted it in Freeform and we wrapped it in NURBS and machined it. It was perfect. Back then (2000), the nurbs wrapping was more manual where you defined the boundaries, but super fast still because you could “feel” your cursor in the valleys where the edges where supposed to go. Since then they’ve done auto-wrap.
@Dan Staples
@ralphg
I was including a scanner in those numbers, just thinking about what it takes to do a complete job, acquisition, post-process, and editing. I’ve done a little laser scan and rebuilding with inferior hardware/software, and it’s brutal. I actually had one project I had to back out of after a week of work (doing a baby doll in SolidWorks from a rigid model). I hate to back out of things, but I just couldn’t do what needed to be done with the tools I had.
I saw an episode of “How It’s Made” on the science channel, and they used I think one of these http://www.creaform3d.com/en/metrology-solutions/portable-3d-scanner-handyscan-3d to scan carvings and then animate them. That episode also had Solid Edge being used to create shipping cases for electronics.
I ought to check that stuff out again. I’m not sure where the line is between Geomagic and Sensable, or if they’ve combined some of the software.
@matt
Geomagic’s latest package, Spark, is $10,000.
@Kevin De Smet
Actually Freeform is totally as easy as it looks. Because you can touch the model (you can feel it through the stylus) someone who has artistic talent will be sculpting good things within a day.
Matt, I think your figures are very high. Back in 2002 it was $18K and I know they worked on driving the price down. You’d have to check with them, but I think it will be more in the $10-20K range (still not cheap, but worth it IMO).
@Rob Rodriguez
Cosmic Blobs ? 😉
@Kevin De Smet
Also, this is from Geomagic which owns Sensable, and he kept referring to the “haptic” which is the 3D stylus on an arm used for sculpting. It was not driven by a mouse.
Great stuff, but between hardware and software, it was probably $40-50k worth of kit.
I’m sure Freeform isn’t as easy to use as it looks, until you spend a lot of hours on it.
the hope that one day they’ll add some catia surfacing and shape&design into V6 is the only reason i haven’t completely lost interest in this product.
in the mean time they really should just buy the geometryworks company and include these features in the base package of solidworks – i’m sure the cost of buying the company would be a rounding error compared to the marketing costs of SWX and would finally make solidworks a serious tool for complex surface creation.
edit: yes, i always look this grumpy as this randomly chosen avatar 😀
@Rob Rodriguez
Yeah, Rob, that’s cool stuff. Wish I could afford that.
@Rob Rodriguez
That’s a cool link, I’ve never seen that software before, thanks.
We share the same view here Matt. I picked up modo 601 @ 40% off last month and am looking forward to learning the ways of the sub-d world!
Great post Matt. I agree the technology needs to merge and it’s taking way to long for this to happen. In terms of controlling organics with engineering precision. Have you seen this?
http://www.geomagic.com/en/products/freeform/demos/
This is one of the reasons why I like Creo. When Creo 1.0 came out they added the freestyle (sub-d) modeling feature to the base package. While you have to start with primitive shapes I can see it getting more advanced in future releases. They also have the flexible modeling which adds a direct editing but I haven’t had a chance to play with that yet.