Deelip.com

Monday, September 17, 2007

CAD Interoperability Problem Solved

Today SYCODE published its first white paper, written by yours truly. In this white paper I discuss how the CAD interoperability problem has been solved. No, I am not high on crack. However, the contents of the white paper are bound to ruffle a few feathers. Here is a teaser:

The Problem
It is a well known fact that CAD software vendors use their proprietary file formats to lock users into using their software. Parametric solid modeling systems make it impossible for their solid models to be worked upon in another parametric system without losing parametric information. They cannot even save to an earlier version of their own software. Users wanting to achieve interoperability between two parametric modeling systems can do so only using neutral file formats such as IGES, STEP, SAT, etc. wherein the solid models come in as dumb solids, making it impossible to edit the parametric features of such models. And this is a big problem. A problem which CAD vendors seem to agree is best left unresolved.

The Solution
The solution to the problem is ....

Download the free white paper here.

posted by Deelip Menezes at 7:49 PM

9 Comments:

  • Might work for basic things you mention like a hole or radius but in reality the parametric information can be more complex than that and you can hardly reverse engineer things like dimensions driven by formulas, etc. There is so much more.

    By Blogger JTB World, At 12:14 AM, September 18, 2007  

  • JTB World,

    It works quite well for all the "basic" things. And more then often the "complicated" things are actually a bunch of "basic" things cobbled up together.

    However, it may happen that some features may be recognized incorrectly or not recognized at all. But the point is that a solution exists, which is far better than the mess that we now have. How well the solution works will depend on the intelligence level of the feature recognition technology employed in each CAD 2.0 system.

    Dimensioning is hardly a solid modeling operation.

    By Blogger Deelip Menezes, At 8:32 AM, September 18, 2007  

  • What I meant was logic like having the hole located 1/3 to the right of the edge. That kind of thing. It's not a solid modeling operation in it self but to have a parametric model that is kind of basic.

    By Blogger JTB World, At 3:55 PM, September 18, 2007  

  • Ah! Sorry, I misunderstood.

    Yes, for that level of intelligence a mere dumb solid SAT, STEP or IGES file will not do. However, such a limitation (and others similar to it) can be oversome by having another "intelligence" file containing this extra information. Quite similar to how a .MTL file stores the material properties and an .OBJ file contains the geometry in the Wavefront Object file format.

    Maybe in the future, a bunch of CAD 2.0 vendors will come together and create some sort of a specification for such a intelligence file. Or they may find a way to hide this intelligence in the neutral file format itself. For example, an IGES file can have geometric data that is perpetually hidden. You could hide the formula as a hidden text object. I know it sounds attrocious, but I think you get the point.

    The good things is that SpaceClaim is a new company that has come out with a new software and has made a concious decision not to add to the prorpietary file format mess.

    Some time back, McNeel came out with their OpenNURBS 3DM file format for Rhinoceros and gave alway the full specification and a library to read and write 3DM files. A 3DM file can hold solid information along with custom data for each object, which can include formulas and just about anything else. If used intelligently 3DM could very well be a foundation stone of a CAD 2.0 world wherein users are not bound by proprietary file formats.

    With the CAD 2.0 style modeling technique, we have the basic framework in place. It probably needs to be improved upon. The CAD vendors that work in that direction are going to finally end up with the market share that they deserve. The same goes for the vendors who work in the opposite direction.

    By Blogger Deelip Menezes, At 6:06 PM, September 18, 2007  

  • Yes, Deelip, CAD 2.0 is exactly the right term.

    Why?

    Web 2.0 is about the “democratization of the Web.” Whereas Web 1.0 was about content being controlled by a relatively small number of content publishers, Web 2.0 is all about enabling everyone to participate. Web 1.0 is worldbookonline.com. Web 2.0 is wikipedia.org, where lots of people can make changes to the entries to make them better.

    CAD 2.0 is not only about the next generation of CAD, it's about the “democratization of CAD.” There are lots of folks that need to work with existing CAD models. Don't make them learn the equivalent of a complex Content Management System. Give them tools that let them edit CAD models as easily as editing an entry on Wikipedia. And just like the way that the democratization of the Web makes content better, the democratization of CAD will make the products we design better.

    CAD 2.0 indeed. This is getting interesting.

    By Anonymous Anonymous, At 7:54 PM, September 21, 2007  

  • There are other CAD programs with these "CAD 2.0" type abilities. IronCAD , Onespace and Keycreator to name a few. This really isn't new.

    By Blogger Kevin, At 1:33 AM, September 22, 2007  

  • IronCAD
    =======
    IronCAD is not quite there yet. True, you can select a fillet face and "change" its radius, but the faces adjoining the fillet face are not changed to match the new fillet radius, and the solid goes all wrong. Basically the feature recognition is localized and does not consider the global effect on the feature on the entire solid.

    OneSpace
    ========
    OneSpace is pretty much in the same boat as IronCAD.

    KeyCreator
    ==========
    I have yet to fiddle around with it.

    Basically, all these programs are on the right track. All they need to do is adjust their feature recognition technology to focussed on the entire model not just the feature being edited. When that is done, they are all set to go.

    In my opinon, SpaceClaim is the only program I have seen where the end user does not miss the feature tree of the parametric solid modeler and can edit the dumb model just as well. True, there may be more that I am not aware of.

    By Blogger Deelip Menezes, At 12:44 PM, September 22, 2007  

  • There is a corresponding need in the AEC space. Here's a comment of mine from a recent post on ArchiCAD Talk comparing ArchiCAD and Revit, and discussing the future of building design and construction in general:

    I feel that BIM apps should be modular, ultimately becoming lightweight clients accessing a shared, web enabled, database. Work sharing across disciplines, and across locations, should be no different than work sharing within the design team. And firms should not be expected to pay on each license for features they never use or are used by only one team member at a time. Beyond that, wouldn't it be great if different developers competed on modules within their field of expertise? A perfect example is rendering. No one likes the built-in rendering of ArchiCAD or Revit, there are many excellent rendering engines out there, and even on a huge project high-end rendering is likely worked on by a single operator.

    Of course this requires a completely different development and business model.

    By Blogger GeoffB, At 1:53 AM, October 06, 2007  

  • "Of course this requires a completely different development and business model."

    Correction. This would need a new CAD software company. The existing ones are pretty much happy with the way things are.

    By Blogger Deelip Menezes, At 12:04 PM, October 06, 2007  

Post a Comment

Subscribe to Post Comments [Atom]



<< Home