Opinions

Re-invention of Direct Modeling

This evening I spend two and a half hours with Matt Carr (Sr. Technical Manager) and Jason Bassi (VP Sales & Marketing) of Kubotek USA. They gave me a demo of KeyCreator 8.5, their direct modeling solution and discussed the “re-invention of direct modeling” as they like to call it. Apparently they were the ones who invented it years before the likes of SpaceClaim appeared on the scene.

This is not a review of KeyCreator. For that I would need to fiddle around with the software, and for that I would need to find the time to do so. But here is are a few things worth noting from the GoToMeeting web conference.

KeyCreator does Direct Modeling, but lacks the push-pull interface. You need to key in parameter values like fillet radius, chamfer length, etc. Kubotek believes that the jazzy push-pull user interface in some of the other direct modeling systems poses significant performance problems for large data sets and is something that is not terribly important. “After all, if you want to apply a fillet you know exactly what the radius should be and you are going to key it in anyways“, said Matt.

From the little that I understood, KeyCreator depends heavily on feature recognition. The software understands features like holes, fillets, chamfers, ribs, pockets and even patterns in dumb solids. I was quite surprised to see that you could even “supress” features, just like you can in history based parametric modeling systems. For example, if a fillet is bothering you, simply supress it and it goes into hiding. Go ahead and model all you want and when you are done simply unsupress it. The feature will come back and sit where it previously was. Of course, there are far more useful things that you can do with feature supression, but I think you get the point.

Apart from Direct Modeling, where you alter the topology of the solid model as you please, KeyCreator also offers Dimension Driven Modeling. Just like in a history based parametric modeling system you create dimensions, which when edited later, alter the model. However, it does not offer a way to modeling by means of formulae and equations. Meaning, you cannot set the height of a box to be twice its length.  I asked if Kubotek was working on it. I got a rather cryptic reply from Jason, “We are not free to disclose what we are currently working on, but I can say that there are no technical barriers that prevent us from offerring this.

KeyCreator is not only a solid modeler. It offers a neat set of tools for surface modeling as well. However, Matt admits, “Our surfaces are not good enough to design modern day cars. They are G1 continuous and in some cases G2, which is pretty much good for most consumer goods.” Another thing, I noticed that in order to model in KeyCreator the model does not need to be watertight. I made Matt remove a couple of faces from the solid model he was creating and asked him to do all kinds of stuff to the remaining faces of the model. As it turns out, even Matt was quite surprised with what KeyCreator could do.

KeyCreator uses the ACIS modeling kernel but has developed a layer over it to do their fancy stuff. There some history to it as well. Three of their top developers had previously helped create the ACIS modeling kernel while they were at Spatial.

KeyCreator comes with complete part and assembly modeling but stores parts and assemblies in a single file format (.CKD). Kubotek also comes with translators for various proprietary file formats like SolidWorks, Inventor, Pro/ENGINEER, UG and (ahem!) CATIA V4 and V5, thanks to 3D InterOp from Spatial, I am assuming with blessings from Dassault Systemes. Sorry, I couldn’t resist.

It appears that Kubotek is more focussed on the geometry creation as opposed to the analysis and post processing. Although they have a 2.5 to 3 axis CAM system, they partner with Algor for FEA, Bunkspeed for advanced rendering, among others. I asked them what they thought about the notion that geometry was a solved problem. Matt said, “Geometry is far from being a solved problem. History based parametric modeling systems may have pretty much stretched their limites, but as far as direct modeling is concerned there is a lot more to be done as far as geometry is concerned.” Actually, I tend to agree. While most of the modeling errors that solid modeling systems spit out may be due to users trying to do the impossible, I believe quite a few errors occur due to improvements waiting to be made to modeling techniques and algorithms. Otherwise people making modeling kernels should have closed shop and gone home right?

I asked Jason about the spread of Kubotek’s customer base. He said that their main presence was in the aerospace and automotive industries. He added, “We have a pretty loyal customer base. Many of our customers resisted the move to history based parametric modeling and are glad that they did so.” Apparently, a few years ago Kubotek did try a few things with the history approach, but soon realized that that was not the right way to go.

I asked Matt and Jason a simple question, “What sells engineering software? Mathematics or Marketing?“. Both agreed that marketing helps bring a product to the attention of a prospect, but ultimately it was mathematics that actually helps a user make a wise decision. Matt quipped, “Mathematics solves customer’s problems, not marketing. And we are interested in solving customer’s problems“. However, they tell me that there was a major restructuring at Kubotek a few months ago which included a new marketing head. So I guess we can expect some more marketing noise from the company from now on. Something tells me that this post is part of that noise. And this brought me to the question about bloggers and their role in CAD software marketing. Kubotek’s opinion about bloggers closely resembles that of Alibre’s. Jason said, “We realize that an increasing number of people are tuning in to what people are saying on blogs and other social networks like Twitter and Facebook. This is a good thing because it levels the playing field. Smaller companies with their limited resources find it difficult to match larger companies that use traditional expensive marketing avenues. The internet changes all that.”

KeyCreator starts at $3700. So I asked Jason if KeyCreator could be considered as a mid-range MCAD system, in the league of SolidWorks, Inventor and Solid Edge which sit at arout $5000. He replied, “We do come close to the mid-range in terms of functionality as well as price. However, we understand that users work in a multi-CAD environment. Instead of locking their data using proprietary formats our direct modeling approach makes it easy for them to exchange data with whoever they need do.” To drive the point in, he opened the CATIA model of a 757 landing gear, edited it as if it was created in KeyCreator itself and saved it back to CATIA. Too bad SolidWorks cannot do that.

A fully functional trial can be downloaded here. However, note that the CATIA translators along with those of Pro/ENGINEER and UG cost extra.