Monday, December 28, 2009

2009 - The Year of … Confusion?

History-based, parametric, history-free, direct, explicit, synchronous, dynamic, instant...

Over the last few years CAD companies have introduced many new terms as they have announced new product capabilities. Most significant are the capabilities related to direct geometry interaction and manipulation. The industry is becoming more familiar with the concept of history-free modeling (direct modeling, direct editing). As the awareness grows, the industry is trying to understand just what this technology means to the product development process. Unfortunately for many it’s still a bit confusing.

Is this history-free direct modeling technology complementary to our parametric history based environment or is it perhaps a replacement? Being part of the CoCreate organization inside of PTC I have witnessed many companies purchasing CoCreate for a variety of reasons. Many have completely replaced their old history-based environment with CoCreate, but many have also added it to their history-based environment as a complementary tool. In either case many companies are seeing significant benefits with history-free direct modeling.

It can certainly be a bit confusing for product development organizations to understand how this history-free technology can bring value and how it fits into the process. The confusion I am referring to is for the most part a result of how many CAD companies are positioning this technology. For example:

  • Siemens introduced Synchronous Technology in 2008. Today they continue to claim: “Synchronous Technology unites parametric and history-free modeling”
  • In 2009 Autodesk gave us a preview of their Fusion technology. They claim that Fusion “unites direct and parametric workflows within a single digital model”

I have had many discussions with people that have been confused by these statements.

Synchronous Technology does in fact unite parametric “control” with history-free modeling, to a certain extent. But unfortunately most people consider parametric modeling synonymous with history-based modeling. Synchronous Technology absolutely does not unite history-based modeling with history-free modeling. The two technologies may sit under a common UI, but they are very separate. Either the user is in a history-based mode, recording every modeling operation -OR- they are in a history-free mode where modeling operations are not recorded. It’s one or the other.

Autodesk does an even better job at confusing the situation. In Autodesk’s statement above they are in fact referring to history-based modeling when they use the term “parametric workflow”. They are also referring to direct modeling in the term “direct workflow”. My first source of confusion is in the concept of a “direct workflow”. There is actually no such thing. Certainly with history-based CAD there is a workflow to consider as every step you take is recorded and will impact the future use of the model. No such concept exists with direct modeling.

The Autodesk statement also suggests the possibility of uniting history-based modeling with history-free modeling with the term “single digital model”. What Autodesk has done with Fusion, a history-free tool (that creates a new file), is to provide technology that can analyze a history-free model coming from Fusion in such a way that the history structure and features of a history-based model, back in Inventor, can be updated based on differences that are found in the history-free model (two separate data models). My primary confusion with this concept is that if in fact the edits are possible in Inventor, then why do I need Fusion? Trying to create intelligent and properly structured history trees from history-free solids is nothing new, but Autodesk has made some progress in this area. I for one am a bit skeptical of this technology as there are just too many assumptions that the system must make to automatically create a useful tree structure, or edit an existing one. Not to mention getting a useful tree structure that would actually match my design intent.

Much of the confusion I see out there comes from the idea that somehow, some way, sometime, history-free modeling and history-based modeling will merge. Recently I watched someone draw two lines on a chart, one representing history-based modeling and one representing history-free modeling. At some point on the drawing the two lines merged into one. The merging of these two technologies is actually not possible. A CAD system is either recording the modeling steps or it is not. It is one or the other. The technologies can coexist, but it must be well understood how the resulting data will be used and managed.

Consider the intellectual property (data) that is created with a CAD tool. If the CAD tool is recording the modeling steps (history-based), the record (history-tree) is the critical IP. Without the history tree, the value of the model is considerably lower. If the CAD tool is not recording the modeling steps (history-free), the 3D model is the critical IP, history trees are of no value. Most any 3D model of reasonable quality can be considered valuable IP with history-free technology. That is not the case with history-based modeling.

While the technology to create history from non-history may improve a little, history-free modeling technology is continuing to improve - a lot. As history-free technology improves the value of the history tree declines. Eventually we may be wondering why we need a history tree, rather than trying to figure out how best to create, structure, modify and manage them.

This post probably does little to clear the confusion, but maybe we can work on that more in 2010.

Have a happy and successful new year!

Paul

22 comments:

Deelip Menezes said...

Paul,

Synchronous Technology (as implemented in Solid Edge) is either this or that. However, Inventor Fusion is history based parametric modeling only. The Fusion part is just the automatic recreation/editing of the history tree while the user makes changes using direct modeling techniques. But underneath a history tree is always there.

I found the way Autodesk implemented Fusion in their Technology Previews (first as a standalone application and later added a plug-in to Inventor) more confusing than the idea of Fusion itself.

I believe PTC is going down the same road with Wildfire 6.0. Just like Pro/ENGINEER, Inventor will always be a history based parametric modeler. Just that the actually book keeping of the history tree can be "outsourced" to Inventor's Change Manager if and when the user so desires.

jonbanquer said...

"Siemens introduced Synchronous Technology in 2008. Today they continue to claim: “Synchronous Technology unites parametric and history-free modeling”"

Am I the only one that remembers Dan Staples head of development for Solid Edge saying that Synchronous Technology would replace history based modeling and that years from now no one would be using history based modeling? Looks to me like Siemens decided that this honest statement that Dan Staples made wasn't good for marketing Siemens NX. That's too bad because I believe it's true.

A combination of direct modeling and history based modeling is at best a stop gap measure until direct modeling tools get strong enough to replace all the history based modeling does.

The reality is that this is now about marketing in the short term and without a doubt PTC is the clear leader in how the have postioned CoCreate and Pro/Engineer. They also appear to be doing a great job using CoCreate technology to attempt to put some new life back into Pro/E.

Jon Banquer
San Diego, CA
www.jonbanquer.wordpress.com

Paul Hamilton said...

Jon, Thanks for the comments. I too agree that there is an end-of-life for history-based modeling sometime out in the future. In an article series I wrote for CADCAMNET back in 2006 I predicted 10 more years. I’ll stick with that, although it could certainly happen sooner.

Deelip, thanks for the additional info on Inventor Fusion, but your comments "confuse" me a little.

Fusion is a standalone history-free system, no history tree underneath – right?

Inventor is a standalone history-based system – right?

Change Manager is a plug-in for Inventor – right?

Change Manager compares the original history-based Inventor model with the imported history-free model coming from Fusion. It then translates the identified changes into suggested edits to the history tree of the original Inventor model. If a solution cannot be represented by editing an existing modeling feature or by adding/removing a modeling feature, the Change Manager will simply recommend the addition of a direct edit to the bottom of the tree.

In the end Change Manager is just trying to make history-based sense out of a history-free model. Not too unlike FeatureWorks. While FeatureWorks adds the history tree and feature information to the imported model, Change Manager modifies the history tree and feature information of the original Inventor model - based on the imported Fusion model. Similar technology, different implementation.

In both cases the systems have to make many assumptions, and in reality way too many assumptions. Maybe that is what you are referring to with "book keeping of the history tree can be outsourced".

As I have mentioned before, any kind of automatic development, editing, management or "book keeping" of a history tree completely dispels the idea that a history tree will somehow support your design intent -- and the value of the history tree continues to decline.

Thanks for your comments.

jonbanquer said...

Paul,

"I too agree that there is an end-of-life for history-based modeling sometime out in the future. In an article series I wrote for CADCAMNET back in 2006 I predicted 10 more years. I’ll stick with that, although it could certainly happen sooner."

To me it's obvious that most CAD companies only look to use Direct Modeling as a way to prop up their History Based modeling tools so they are not seen as obsolete technology. This is only going to work until Direct Modeling gets stronger and more users get the clues they don't have now.

In my field, machinists are overjoyed with SolidWorks because they have had to use the garbage CAD that's in Mastercam, Surfcam, Smartcam, Gibbscam, Edgecam, etc. for years and they are more than willing to ignore the massive problems that SolidWorks has with non-native solids as well as SolidWorks total inability to easily let someone make major modifications to models they didn't originally design themselves.

One day an independent CAM company with a clue is going realize that in order to differentiate themselves they will need to run inside a quality direct modeler. If that solution is marketed correctly that CAM company will have a significant advantage over all the CAM products now lining up to run inside of SolidWorks.

To me it's sad that the CAM industry is mostly an industry of followers rather than leaders. Not sure how U.S. manufacturing can ever get their act together and complete on a global scale with the current prevailing mentality.

Thanks for letting me comment here, Paul. There isn't a chance in hell that my comments would get published on most other blogs run by CADCAM software fanboi's and by those who's real agenda is to sell as many different software add-ins as possible.

Jon Banquer
San Diego, CA
www.jonbanquer.wordpress.com

Deelip Menezes said...

Paul,

In the next technology preview (hopefully), the standalone Fusion application will no longer exist. I presume there will be only be Inventor with the Change Manager built into it. This is the confusion I was talking about when I referred to the way Autodesk packaged Fusion. Now they have three separate things (Inventor, Fusion and Change Manager) and you really need to look at them in the bigger picture.

I also believe that the bad packaging is also the reason for the Change Manager failing to arrive at the correct solution. During my tests I found that the Change Manager did a good job if I let it run after every change I made. But if I made multiple changes and then asked it to figure it out, it usually did not get it right. I guess this is what you meant by the technology requiring to make "many assumptions".

The Change Manager, when put to work after every change (which I think should happen), will work very differently from FeatureWorks. In FeatureWorks you give it a dumb model as ask it figure it out. That's trying to solve one huge problem. If the Change Manager is made to figure a tiny bit of the huge problem every time, given that it has access to an intelligent parametric model as a high quality start point, I think the results will be far better.

I guess we will have to wait for the next Technology Preview to see what Autodesk actually does this time.

Mike said...

I'm still not convinced that history free modeling will ever be anything more than a niche technology. At the most it will be an add in like NX is doing. There is just too much advantage to having that history especially in parts that have several different "phases". Our company is dropping CoCreate after using it for several years because it couldn't handle edits on complex castings. We have a guy work on a for a couple days and after the design review he'd have to start over because the fillets in a certain area were too hard for the software to edit. I watched that happen several times a week for the past 2 years. We knew that it was time to switch before we had too much 3D data invested in a software we didn't think could handle the task at hand. Until these direct editors get better at editing their own data they'll never be more than a novelty.

peak-limits said...
This comment has been removed by the author.
peak-limits said...

Mike
I completely agree with you, the history based design is more required when you are making complex designs. The history structure helps to study the process of designing by following the series of features being created.
SolidEdge had a good concept but is now quit confusing to work with. I think it is still possible to have a history based modeler, combining with the concept where you don't require to edit sketch to modify the model.
So think about a CAD in which you make a cylinder out of a sketched circle and when you modify the radius in 3D, it automatically updates in 2D sketch of circle and all the details from the sketch to all the features are saved in a history structure.

So till now we were only able to modify 2D sketch to modify 3D model, now the CAD should have the capability to update 2D from 3D

Paul Hamilton said...

Here we go again – another “which is better and why” debate. Thanks for the comments though. It keeps us thinking – and hopefully progressing.

Mike, I can agree with you that there are editing challenges with history-free modeling especially with complex castings. There are some simple techniques that you can use to work through these challenges and continue the design process with the existing model. You should never need to “start over” in “any” history-free system. I would be glad to show you, although it sounds like you have already made your decision. What was the latest version of CoCreate that was being used?

Most all of the enquiries PTC gets for CoCreate Modeling come from companies that are long-time history-based users. The number one reason they come is due to the complexities and sometimes inability to make the necessary edits to their history-based parts – especially complex castings. Examples of being forced to recreate the model are all too frequent.

Both tools have their editing challenges. With history-free you work through the difficulties by adding and/or removing faces in the problem area – usually a vertex region of a blend. With history-based you attempt to fix the history tree or in many cases just “start over”. I hope that there was more to your decision criteria than just the ability to edit geometry. If not your company is wasting a lot of money.

With each new release of CoCreate Modeling, as well as Spaceclaim and other history-free tools, the editing challenges are reduced. There is no technical reason that this trend won’t continue. In a history-based system the only way to reduce or eliminate editing challenges is to design and validate your part – before you model it. (Some people call that “planning ahead”.)

Peak-limits, I just have one question about your comment – when and how does a history tree ever accurately represent the design process? I understand that if you’re really good at modeling, the tree might represent the modeling process, but don’t see the value in that. If the tree could represent the design process, you could have designed your part in Pro/E, SolidWorks, Inventor, Solid Edge or whatever H-B system you want and the history trees would be identical. History trees are too tightly tied to a particular CAD systems capabilities and available modeling functions to be useful for representing even a part of the design process.

OK – two questions. If you allow the system to change your history tree (sketch) based on a direct edit (modifying the radius of a cylinder) how could someone possibly study your history tree and come to any conclusions about your design process – or design intent?

Capturing the design process is best done in a notebook and/or a PDM/PLM system. To think otherwise is very risky.

Mike said...

Actually there were several reasons for the switch. I mentioned geometry frustration because that was a personal pet peeve of mine. If I had to list them they would probably be; geometry creation difficulties, lack of native support with CAM software, lack of a robust integrated FEA/CFD solution, weak PDM, difficulty finding users already trained before hire (Us: "We use CoCreate here." Them: "What the hell is that?"), the final straw though would have to have been corporate making a push for "global engineering". We looked at 3 other systems but went with SolidWorks in the end.

We ruled out Autodesk because their PDM was no better than what we had.ProE would have worked but by the time we added in all the functionality we wanted we might as well buy CATIA because of the cost. SolidWorks was the best choice we felt because it was a cost effective system and corporate wide SW seats out numbered all other systems 10:1.

Anonymous said...

two questions:

how much of PTC's R&D $$ is being spent specifically on cocreate? And of that R&D $$, how much is being spent improving the core functionality vs. just creating interfaces that allow PTC sales to try to sell the rest of PTC portfolio to cocreate customer base?

Paul Hamilton said...

Mike, thanks for the additional info. PTC is helping to get the visibility up on the CoCreate products, but there is much more to do for sure.

I am not sure how much money PTC is putting into the CoCreate product line but the next release due out this spring is one of the more significant releases in many years. Not only in robustness, predictability, usability and core capabilities, but also in the areas you mention. By integrating PTCs Granite interoperability kernel into CoCreate, we now have associativity with Pro/E NC, Pro/TOOLMAKER, Pro/E Mechanica, IsoDraw and many other complementary products. They are not integrated under the UI, but it's a good start.

Some may view that as an attempt to provide a more complete tool set to the CoCreate user base. Some may view that as an attempt by PTC to grow their business. Both would be good.

Anonymous said...

It is my hope that PTC will provide more transparency into cocreate R&D investment.

Kubotek Creation Engineering said...

My prediction is that this will be a year when many companies using history-based parametric modeling will discover the benefits of 3D Direct Modeling, which by definition is history free.

Kubotek's KeyCreator 3D Direct CAD has been a pioneer in this technology. Spaceclaim is the new comer has given it a sexy user interface although they lack the depth or maturity that you get from a widely used and well established CAD tools like CoCreate and KeyCreator.

I also agree that the marriage of Direct Modeling and CAM makes perfect sense. De-featuring the model is a snap with 3D Direct(history- free) CAD.

I agree that parametric history based modelers will decline and eventually be marginalized; The time frame may be somewhat less than what Paul has predicted.

History-based parametric programs were developed at a time of slow processor speeds and smaller amounts of memory. Thus, the need to create a recipe of how the model was created was then required to create large and complex models with the accompanying large files.

With today's faster processing speeds and huge amounts of memory, 3D Direct Modelers can perform complex modeling tasks in realtime, (KeyCreator can actually find features and patterns of features in a dumb solid (non-native file) and change them in realtime). And without all of the overhead of the history, the CAD files are much smaller and easier to manage (Which IT departments love.)... but we wont get into that today.

Thank you for the thoughtful discussion.

Scott Sweeney
KubotekUSA

Paul Hamilton said...

Thanks for dropping in on the discussion Scott. You make some good observations and points.

coroto.tgb said...

Maybe someone who has used history-based modeling extensively can help me understand this better. I can see how the value of the intellectual property embodied in a completed 3d model is there in the ultimately realized 3d geometry. I can see the value in the associated dimensions, tolerances and parametric relations that drove that final 3d shape. But I'm having a harder time getting a handle on the value of the history-tree per se. Apart from the final 3d geometry and the dimensions/tolerances/parametrics that created that result, what is the unique value of the IP of the recipe, the history of the modeling steps? When is the history itself significant?

Jaybassi said...

Jon - When you come across a CAM company who wants to run inside of a direct modeler, feel free to direct them to Kubotek. We have thought this for many years, but have not found any takers.

Paul Hamilton said...

Last week I received some questions via email from a reader regarding this post. I asked his permission to post his questions and comments. Here are his two questions and my responses (with some edits):

“how is history-based modeling superior to history-free?”

My response: As history-free modeling improves, history-based modeling becomes less superior. Today I actually think that “in general” history-free technology is edging out history-based in its ability to support the bigger picture of product development (although some H-F tools are very immature and narrow – the technology is here). At the current state of these two technologies it is now important to understand the characteristics of the product that you are trying to design, the product development process, and the product life cycle, to really make a good decision on which fits best. One of the last remaining benefits that history-based brings over history-free is in freeform surface design and manipulation. (I discuss this in today’s post "Direct Modeling and Freeform Surfaces")

“Is there any reason why a green-field organization (with no legacy data) would start with a history-based CAD solution?”

My response: There are several reasons that I will list below – BUT there are very few technical, or capability related reasons. Here are some of the reasons people purchase history-based systems today:

• They don’t know that there is an alternative
• The people they are hiring already have experience with history-based and change is considered difficult and expensive
• History-free implies that something is missing, a risk • For many history-free is something new, another risk • Everyone else uses history-based successfully, we can too
• Many people, especially IT, don’t understand the long-term impact to process – CAD is just a tool, let the users decide.

Considering the state of these two technologies, I think it is a big mistake for a “green-field organization” to assume that history-based is best. History-trees are proprietary data with a very short life span and can only be manipulated by a handful of CAD users, (relative to the broader set of people that could get value from CAD data). I don’t think people yet understand how much wasted effort goes into creating and managing them. I consider it “wasted effort” because there is now a very capable alternative. Hope this helps a little.

His response (and the reason that I am posting this): “It does. My interest is in manipulating the data outside the CAD system – i.e. doing something useful with variants or storage. History-based modeling makes that a horrible process – So, I’m kind of waiting to see when history-free modeling becomes dominant and we can do more interesting things with the downstream data. IMHO: CAD people always seem blinkered with what can be done with all that data that gets generated.”
-----

Do we really understand the complexities and costs related to the management of a history tree?

jonbanquer said...

Scott,

You wrote:

"Spaceclaim is the new comer has given it a sexy user interface although they lack the depth or maturity that you get from a widely used and well established CAD tools like CoCreate and KeyCreator."

I agree and what's really pathetic is no one will stand up and say how badly SpaceClaim handles something as basic as holes.

If SpaceClaim can't easily handle something as basic as holes I don't feel they should be spending one minute of time on total nonsense like multi-touch screen technology.

Don't get me wrong, I love SpaceClaims Push / Pull technology and KeyCreator should adopt it but if SpaceClaim can't handle basic mechanical stuff like having a hole wizard or hole features then Push / Pull soon amounts to yet another CAD modeler that can't do the job.

When is Kubotek going to start playing hardball and lay it all out (like I've just done with Spaceclaims inability to easily edit / manage holes) so potential users know what the downside is?

Why should I have the only blog that's willing to play hardball and lay what the real deal is out on the table?

PTC has stepped up and has started to market Direct Modeling like I've always believed it should be marketed. How soon before Kubotek takes off the gloves and starts kicking some donkey by exposing the dirty little secrets and dirty little lies of its competitors?

Jon Banquer
San Diego, CA
www.jonbanquer.wordpress.com

jonbanquer said...

Jay,

You wrote:

"Jon - When you come across a CAM company who wants to run inside of a direct modeler, feel free to direct them to Kubotek. We have thought this for many years, but have not found any takers."

They aren't going to come crawling to you begging for help even though they should. You need to be banging on their door letting them know the answer isn't just running inside of SolidWorks and that they need to get their stand-alone applications up to speed before their investment in their stand-alone CAM becomes worthless.

You want a name of a major CAM company owner who will give you the time of day? E-mail me. I'll give you his name.

Jon Banquer
San Diego, CA
www.jonbanquer.wordpress.com

Jaybassi said...

Jon,

You want a name of a major CAM company owner who will give you the time of day? E-mail me. I'll give you his name.


What is your email address?

jbassi@kubotekusa.com

Nick Sidorenko said...

Hi, Paul

May be, I’m little bit late with my comment, but let me still say that all the discussion around “history-based” and “history-free” seems to me misleading.

The discussion should be around “direct modeling” and “parametric modeling”.
I know that for somebody these are the synonyms - “direct modeling” is “history-free”, and “parametric modeling” is “history-based”. But, actually, they are not.

History tree is definitely the evil. It has to be thrown away from the CAD life and I do believe that in the nearest years it will become atavism. But I can’t say the same about parametric modeling.

Parametric modeling was the most remarkable invention of CAD Revolution initiated by PTC 25 years ago.

The implementation of this modeling, using “feature-based” (or in another words “history-based”) approach was questionable (though, frankly speaking, at that times there were no other way out). All the talks that history tree expresses the “design intention” are from the devil.

The reality is in the fact, that those methods, which are implemented in modern parametric solvers, are not able to resolve complicated parts. In order to allow designers to work with the complex parts, the parts are divided into numerous “features”, and these “features” are consequently resolved by these poor solvers (to be more precise, the systems of constrains equations are solved in corresponding 2D sections).

Were we have a solver that is powerful enough to resolve the systems of constrains equations for the whole part (when the part is complex enough, and these constrains are 3D constrains, not the 2D constrains as in the traditional parametric CAD) there would be no need in the feature-based (i.e. history-based) approach.

So, the root of the problem is in the computational difficulties of the archaic methods used in traditional solvers. All the next is the philosophy fabricated around this fact in order to hide this sad reality from the users.

And now let’s go from the sad reality to the bright future.
During these 25 years the mathematics was not standing on the same place. New remarkable methods were developed to solve very fast and with a very low memory requirements the systems of equations that are typical for parametric CAD (both 2D and 3D). And these methods are already implementing in the new generation of the CAD software . This new solid modeling CAD will unify in a common workspace all the direct modeling features and the full range parametrics in 3D space.

When such an application will be implemented (and after some years it will become the standard for the industry) there will be no place for discussions – “direct modeling or parametric modeling”. Use what is more convenient for you at the current moment.

We are already working in such a way with 2D parametric sketchers – we can drag and drop some entities (these are direct modeling on the 2D plane) or we can change some dimensions and solver automatically regenerates the model (this is parametric modeling).

So, what we are to do, is to repeat this paradigm in 3D space. And this is not a dream – it is the reality of the nearest two years.