Fusion Discussion On Twitter

By Dmitry Ushakov (@dmitryushakov)

CEO – LEDAS (www.ledas.com)

The recent announcement of Autodesk 2012 products was intensively discussed on Twitter. My attention was attracted by an interesting discussion about Autodesk’s decision to make Inventor Fusion a part of AutoCAD 2012.

The discussion was started by Blake Courter (@bcourter), a founder of SpaceClaim and ideologist of direct modeling. “Thank you Autodesk!” he wrote. “For what?” asked Deelip Menezes (@DeelipMenezes). “Not making Fusion an individual product? 😉” Deelip meant that Fusion is a competition solution for SpaceClaim. But Blake explained, “For trying to make direct modeling the most prevalent form of solid modeling. And I can’t wait for it to go standalone.  Imagine growing up with an iPad and never using a laptop. Now imagine growing up with direct modeling and trying to move to history- based.

In what followed, Deelip and Blake agreed that the iPad cannot replace a laptop in some things. And neither could direct modeling completely replace history-based approach (albeit the latter is sometimes “awkward and convoluted”). But “the  iPad is a version two. SpaceClaim is at version seven. [What will be] In ten years?” wrote Blake. “I suspect my children will never experience regeneration failure,” he added. Later Blake explained to me that he does not have children yet, but hopes to have someday. This raises doubts on his prediction ;-). But he gave at least 20 years to history-based system to retire.

Then Deelip Menezes made an important statement that gave a new impulse to the discussion: “Solid modeling should be more about the thing being modeled and not how you are modeling it.” It attracted attention of Sean Dotson (@SeanDotson), President of RND Automation, a company that engineers, designs, and manufactures custom automation & material handing machinery, including robotic work cells. RND, a longstanding Autodesk customer, was recently named by the vendor as “Inventor of the Year”, which means that Autodesk recognized their design and manufacturing processes as the best practices in the industry. Sean answered to Deelip: “So you are saying the ends justify the means? Nah. I’ve seen some great models that were constructed very badly.” Here I can add that Sean is not the only person who has seen such models ;-). But Deelip refined his statement: “Simply saying that too much time goes in planning how a model needs to be created.

Indeed, it is very important to make the things simpler, Sean agreed. But “as long as we do not loose the design intent.” Sean needs to be able to know why the part was designed the way it was. A lump without history tells him nothing. He gave an example: “a hole in the middle of a 2″ wide plate. Was the intent to be centered or 1″ from an edge?

That example doesn’t need history. Constraints are one way, with varying strengths and weaknesses,” Blake answered. Gold words! Three years ago I wrote a paper “Variational Direct Modeling: How to Keep Design Intent In History-Free CAD”, where the idea of using constraints for editing “dumb” solid geometry was presented. It is a new application of geometric and dimensional constraints, which are now widely used in 2D drawing/sketching and 3D assembly design. But 3D geometric modeling is much more complex! Hopefully recent evolution of solid modeling kernels (like ACIS by Spatial) allowed one to easily update boundary model with the results of constraint solving procedure by just moving/deforming its faces and their boundaries. And the main issue here is to provide user with the solution he/she expects. In other word, the intelligent direct modeling system should be able to recognize and store the design intent within any “dumb” geometry and take it into account when the user changes the model. Constraints serve as a universal language for expression of the design intent here. They are easy to recognize (much more easily than features), easy to keep (thanks to constraint solving technology) and easy to add to existing geometry (you don’t need to reconstruct the model).

Concerning the example given by Sean, I tweeted that an intelligent application should recognize several possible design intents in one model and propose the user to choose the one he/she prefers. Moreover, with constraints the user can express any desired behavior, as in this video recorded a year ago, when LEDAS started to work on the implementation of Variational Direct Modeling in Rhino:

At the end of the discussion Blake Courter came back to the question he initiated it: “Who can give me an estimate of how many Fusion seats will be installed in the immediate future? Anyone think there will be fewer than 1M seats of Fusion issued? That’s about the number of history-based seats under maintenance.

I visited SpaceClaim.com to find more information about the Blake mentioned. Here is the citation:

It is estimated that the penetration of traditional history-based 3D CAD is very low among more than 15 million engineers involved in manufacturing worldwide, with fewer than one million estimated 3D users in the $5 billion mechanical CAD (MCAD) market.

This estimation seems strange to me, since SolidWorks recently reported 1.4 M installations, and there are some other 3D MCAD systems on the market ;-). On the other hand, we now have Inventor Fusion as a part of AutoCAD (and 1M seats are absolutely reachable in one year), and also SketchUp, Rhino, KeyCreator, CoCreate (Creo Elements/Direct), which total number of seat adds another 1M to the direct modeling pan.

David Levin (@levin_david), LEDAS founder and chairman, noted “Mainstream AutoCAD users will hardly be infected by Direct Modeling. Moreover, I am not sure whether such users will soon go beyond 2D. Finally, I think that a concrete mixed approach of Autodesk may (compared with automatic detection of the design intent) be not optimal for overall global development of Direct Modeling. However my skepticism does not abolish or downgrade significance of the move just made by Autodesk, be it only marketing or not.”

In the neart future, I believe Direct Modeling systems will prevail over history based ones, or at least be compariable. But what’s next? Will the history-based approach be retired? Or will both branches of the 3D CAD tree bear fruits? Can we use any analogy with the iPad/laptop market here? I don’t know the answers to these questions, but expect very interesting and hot discussion at COFES 2011. And of course I’m going to take part in it. 😉

About the Author

Dmitry Ushakov is the CEO of LEDAS Ltd., a Russian-based software development company that provides component technologies and related services for the CAD/CAM/CAE market. He was hired in 1999 by LEDAS to be the lead programmer, and then was promoted to the position of chief technology officer, and later director of product management. During this time, he initiated and supervised the development of LGS, the LEDAS geometric constraint solver, which defined the LEDAS business model for successive years. He is the visionary for a new variational direct modeling technology for the next generation of CAD, as implemented in the series of Driving Dimensions plug-ins for 3D modeling software. In 2011, he was appointed Chief Executive Officer of LEDAS.

  • Anonymous

    Where was I for all that discussion – must have been deep in something else? Anyway- here are my observations –

    1st – Levin from Ledas is absolutely correct, once AutoCAD users experience the simplicity and power of 3D Direct modeling – there will be no reason to ever use history based modeling.

    2nd – History -based modeling may have a few more years left in it, but it will have a steep decline in the next decade. 20 years from now – it will be used by only the old guys that grew up on history-based modeling. It’s difficult for history based CAD users to understand that you don’t need history to convey design intent.

    3rd I thought Fusion was intended to be a fusion of History -based and Direct (which I still don’t understand – you either have history or your don’t.) I think Fusion is ConFusion!

    My 3 cents,

    Scott Sweeney

    • ushakov

      Concerning the “ConFusion”, I agree. I also don’t like this kind of mix of direct and history approaches (that can be found also in Solid Edge ST3), since it makes simple things much complicated. But I suppose that such hybridization has a sense for many users – just like a hybrid vehicle is an intermediate step from petrol to electric engines.

  • Dnelson

    I think Blake is slightly over optimistic about people in Autocad using fusion as a stepping stone to 3D modeling. This is what 3D solids in Autocad were suppose to do in the 80’s, and Autocad today still mostly used as 2d drafting program. Especially in the AEC market where I work, firms don’t even touch the 3D environment.

    • ushakov

      The main difference between 3D solid modeling in AutoCAD 2011 and AutoCAD 2012 (with Fusion on board) is the ability to EDIT your 3D model. I believe it will change the things drastically.

  • Joe

    It’s a little fishy that the people representing direct modeling programs tout that they will control the industry. History based modelling will always have a place in the industry. It may lose market share, but both methods have a place in design. To say it will go away is bombastic.

    • Nainar

      Right on dot mate!! It will be users choice and I still dont know how one can build a complex model such as an aircraft or a yacht totally using Direct modeling. From my understanding, Direct modeling will come in very handy in late editing of model which is a huge problem with History based modeler, especially when one has to do those with manufacturing considerations. I go back to my previous company Think3, which had direct modeling coupled with history!Yes, you can rotate a face by 5 degrees to give a draft and it will be replayed when the model gets rebuilt through history based data!!I feel that is the place Direct Modeling will fit.For those predicting history based modeling will go away 20 years from now, take a reality check. Some of the FEA solvers are still in Fortran. It is just that we dont know, that will be a modest prediction.

  • chad

    I don’t see direct modeling being the only reason to move into 3D. If your not using 3D now then you’ve already missed the boat. 3D is only a portal to achieving something greater. The future is going to be about how you leverage your 3D data not necessarily about how you work with your 3D data. direct modeling is getting way over hyped.

  • David Levin

    Deelip, thank you for quoting me. Yes, its 1st sentence “Mainstream AutoCAD users will hardly be infected by Direct Modeling” exactly corresponds to my thought. But I can’t semantically match the rest of citation with what I wrote in http://bit.ly/eI7H13 . 🙂

  • Guru

    Wow.. Loved the article Mr. Dmitry Ushakov.. just becoz you gave us the summarized edition of a global twitter conversation happening between great mind and on a product launch.. which in itself, I feel is a unique way to present some great facts to readers… Twitter is becoming a great tool to kick in meaningful conversations… thanks for the writeup… 🙂

  • Proe-warsztat

    there is serious doubt about leaving by feature based models the scene of CAD modeling approaches within several incoming years for good. This funcionality is simply required when it comes to have accurate control over the things which are developed.

    But…, since it is tedious job to play with FB models, direct modeling will make its way ahead and spread quickly across wide range of CAD platforms for sure, because social factor of making things happen requires to allow no CAD users like managers to play with geometry within simple interface.

    To me, it would become even more rapidly if there will be available, stable process to compute or translate geometry to feature based model. I can imagine that on prototype phase of project fillfulled with great amount of meetings accross different deps. and location people are sitting with its ipads, which are used as control device only reflecting the simplified interface of displayed CAD software, and playing with a model, by moving this or that. At the end there is a moment for experienced engineer to polish the model. This should be the phase to come back to feature based modeling.

    • ushakov

      Features will not leave the scene. They will exist as declarative features. Based on constraints, not on the history tree. Feature modification will act locally and won’t require the regeneration of geometry.

  • A worthwhile comparison could be how you write code for an application. What is the need ? What is the long term use ? Are newer engineers expected to come in and maintain the application and perhaps enhance it ?

    An application can achieve the same result whether internally written badly, written in assembler, written in C, C++, C#, language du jour or just as an EXE.

    What matters in the long term is how much of the intent (hence IP) is being captured in the investment (source code) and whether there is an envisaged need for capturing that intent.

    To me, feature recognition for direct editing is a bit like writing an app in C#, compiling to an EXE, then disassembling it back into IL (intermediate language), making a change and recompiling it back into an EXE. 🙂

  • Proe-warsztat

    and this is the moment when or where tires hit the road. Programing environment is good example. As one collegue of mine told me – if You program in C, and try to come back to Assembler, it means You did something really wrong while writing the code. That is true, however there are moments of optimalization where simply C is out of scope of interest.

    That is never ending discussion which tends at end to minor conclusions. To me it looks like development of different CAD apps towards to common point. Apps like Rhino expand its capablities over FB funcionality, in the same time Pro/E, SW, etc, researches the tools to faciliate control over geometry.

  • How about the best of both -> truly merging histroy and direct in a single model = http://www.ironcad.com

  • Jeuron Winser

    Speaking of “will both branches of the 3D CAD tree bear fruits?”, you can check a new challenger, ZW3D 2011, to be released soon I guess, here is their short video on youtube: http://www.youtube.com/watch?v=QWtCkZgyxus

    but they called it “direct edit” instead of “direct modeling”, fishy~