Monday, February 9, 2009

Inventor Fusion – More ConFusion

I'm pulling a few quotes for an article by Kenneth Wong published at Desktop Engineering.  The title of the article is: “Autodesk Joins the Hybrid CAD Movement with Inventor Fusion".  Has Autodesk solved the unsolvable?

Comment #1:

So what distinguishes Inventor Fusion’s underlying technology from its rival Synchronous Technology, made available via Siemens’ NX and Solid Edge software? The answer is, in Anagnost’s words, “a two-way street; the ability to move back and forth seamlessly between the two modeling paradigms.”

When creating 3D geometry, a CAD system is either capturing information about the modeling operation; history-based, or it is not; history-free.  When editing, you are either editing the captured information resulting in different geometry, or you are directly editing the geometry.  It really is that simple.  Some argue with me on this, but how can it capture and not capture at the same time?  It can get a bit more complicated with direct edits within a history-based system.  These edits (including information about the edit) are usually captured within the tree to preserve them during the next regeneration of the tree.  The exception is when a parameter of an existing modeling feature, within the tree, is dynamically and graphically changed and it appears to the eye as a “direct edit”.  In any case the rule still applies. So how do you “move back and forth seamlessly between the two modeling paradigms” of creating history and not creating history, or perhaps editing history or editing geometry?  Hum... Sounds confusing to me. 

Comment #2:

Anagnost says, with Inventor Fusion, you can make a change using direct editing, but if you wish to parametrically refine the geometry deformation after the fact, you may go back to the feature tree to modify the parameters. He clarifies that Inventor Fusion automatically figures out the parametric modeling steps necessary to accomplish the direct edits, “including the order of dependencies.”

This comment only refers to editing geometry, so maybe I'm making the incorrect assumption that you can actually create geometry/parts using Inventor Fusion.  Maybe it is just an editing tool.  The comment: “Inventor Fusion automatically figures out the parametric modeling steps necessary to accomplish the direct edits, “including the order of dependencies.”” would indicate that Autodesk has solved a very complex problem that many companies for many years, including SolidWorks, have been trying to solve: the ability to scan a dumb solid, or partially dumb solid, and build a useful feature tree, including parameters, that fits the users design intent.  This statement would indicate that Inventor Fusion has the capability to walk through the topology of a solid model and identify and recognize meaningful features from it (this is nothing new: feature recognition). Then it would be able to create and associate sketches (if needed), parameters and conditions to these recognized features. Then, order the features in a meaningful and useful way.  And it would all need to be done in a way that would fit the users design intent - “automatically”.  This would be very cool, but I kind of doubt it – show me.

Comment #3

"Dan Staples, director of Solid Edge development at Siemens, says, “If you make a decision to move to Synchronous Technology’s way of modeling, you actually have no reason to revert back to the traditional way of modeling. Synchronous Technology is a superset of technologies.”"

When you move from “History Mode” to “History Free Mode” in NX or Solid Edge, you lose the history tree, and the system will clearly warn you of this.  With Synchronous Technology you can still parametrically control the history-free model, ("superset of technologies", i.e. parametrics with history-free modeling), just like you can in CoCreate Modeling, but you do lose the tree.  With ST you can toggle back to “History Mode”, but it will not and cannot recreate a useful tree and feature structure from the history-free “dumb” model.  As such there is infact, as Dan states, "no reason to revert back”.  The term "superset of technologies" is refering to the combination of parametrics with history-free modeling, not history-based with history-free.  Combining parametrics with history-free modeling is something that most all history-free systems on the market can already do.  It's nothing new, although ST does a nice job of it. Are these guys talking about the same thing?  "two modeling paradigms"? "superset of technologies"?  I'm confused.

It is great to see Autodesk jumping on the direct modeling wagon to further validate the need for more intuitive and flexible modeling tools, and it will certainly be enjoyable to watch as things mature.  Maybe they really have invented some amazing new technology.  We will watch and see.  Just don’t get too excited or perhaps baffled by all the flashy marketing stuff.

Paul

8 comments:

Jeff Waters said...

Another good post, Paul. I'm with you on this one. I also don't really believe there is such a thing as hybrid history/history-free modeling. This is simply a marketing ploy.

I'm not sure everyone is going to take the time to follow what you are saying, but you are correct... to move from history-free to history-based would require a tool to look at some arbitrary piece of geometry and magically create the underlying history-based recipe that could make such an object.

AND it would have to take into account the personality fo the individual designer and the design standards and rules of that particular company (there are many ways to skin a cat... and many ways to draw the same history-based part).

We need a good analogy for this marketing hype. Sounds like something a horse-drawn buggy company might try when announcing their first motorized vehicle. I can picture a Model-T dragging a near-dead horse: "Our hybrid system let's you choose!"

R. Paul Waddington said...

Paul,
Autodesk have had a form of direct editing, of 3D objects, in AutoCAD for some time. Very limited functionality, have used it on a number of occasions to check out its capabilities, show those interested and to give ‘some’ a feel for what the much more capable ST tech’, and others can do. Does show tho’ Autodesk have been playing with ‘an idea’ for a longer period than some may think; in a product, Autodesk, it would appear, would rather not be used for 3D as evidenced by the dearth of discussion and information.
One major marketing stumbling block for Autodesk will be the fact they used Inventor to spit on the effort of a lot of early 3D AutoCAD/MDT guys and those being quoted now are the same who quite openly said capabilities of their own product(s) COULD NOT go forward and functionality like ‘ST’ and ‘Fusion’ was simply NOT possible. Oh, how a few years and some competition can change people views.
Your comments about ‘flashy marketing stuff’ are very timely words: firstly because many of us remember what Autodesk have said and done in the past and secondly because Inventor was a product that defined Autodesk, and its developers, as market followers and followers need to buy room to maneuver. The press about Fusion is about ‘muddying the waters’; Autodesk are using the same techniques MS used, to kill a very good OS, before releasing Windows; and both Windows and Inventor, when released were not up to expectations and some would say Inventor still is not!
I’m with you; its good to see some action, disappointed tho’ when look when looked at in the light of the knowledge of the opportunities that existed a loooong time ago.
Fusion is going to need to be such a stunning step forward to be of value to Autodesk’s customers; if its introduction/capabilities mirror that of Inventor it will create huge disillusionment in Autodesk and those looking to stay in the Autodesk fold and move forward.

jonbanquer said...

"When you move from “History Mode” to “History Free Mode” in NX or Solid Edge, you lose the history tree, and the system will clearly warn you of this."

So don't move to "History Free Mode". You only gain two ST tools when going from "History Mode" to "History Free Mode" in NX 6

1. Shelling

2. Bisector

Stay in "History Mode" and accept that for now you lose the use of only two of the many ST tools. By doing this you don't lose the models history.

The changes we would need to make for the models we work on don't require us to Shell or use the Bisector tool. Is this a "perfect" solution? No it isn't but it's the first release of ST in NX and I accept that these limitations will most likely soon be removed.

Jon Banquer
San Diego, CA
http://jonbanquer.wordpress.com/

jonbanquer said...

Inventor-Fusion is little more than vaporware.

For well over ten years SolidWorks Corp. has been obsessed with Autodesk. My bet is that SolidWorks goes the exact same route as Inventor-Fusion. Both SolidWorks and Autodesk are very badly behind when it comes to Direct Modeling tools. Even so, if this is the road that it takes to get more Direct Modeling tools into the hands of more users than so be it.

As time goes on and the Direct Modeling tools and user interfaces get better you will have more users that don't want to use history based modeling tools and just want Direct Modeling tools.

Jon Banquer
San Diego, CA
http://jonbanquer.wordpress.com/

Mike said...

The way I figure it, you'll never lose history completly nor should you want to in most cases. There are only two or three cases that I see where true history free approach is the optimal solution. Injection Molded parts, parts with no secondary operations once they are first cast/forged/machined and parts that are being machined from billets. With a weldment you really don't have any need to change profiles, just lengths and angles. With parts machined out of castings/forgings you want to have a model that updates common features between the two when you update one or you're doing your work twice.


The only good thing that a history free system gives you is the ability to do some direct editing and from what I can tell that's sketchy at best when you start dealing with a complicated part. If you have a lofted feature with blends and draft moving that feature is a bit dicey from my experiance or of you want to change the the location of one of your profiles in a loft that's pretty tough without blowing it all way and starting again.

A lot of you guys think that "hybrid modeling" is a marketing gimick and for the most part I'd agree with you but I also think that history free is a gimick with an even smaller niche market.

R. Paul Waddington said...

Mike said, “The way I figure it, you'll never lose history completely nor should you want to in most cases.”

At the risk of being hammered I am going to, in part, agree, with Mike’s statement but at the same I’m going to go out on a limb stating; no one modeling (or 2D vs 3D) method is better, or worse, than another except in the perception, and application, of a particular method (or product) as applied by an individual for a particular task or personal preference. Gasp!

Nothing focuses argument in the CADD world as quickly as a comment about 3D being better than 2D and History better than Non, etc. None of the arguments hold true for all. The problem being, each contributor tends to use their own experience or requirement(s) as their reference point; that is entirely understandable but, we all have to think more broadly about the wider effect CADD has on the development of our industries and on individuals.

CADD vendors tend to narrowly focus too: they may argue, because, they are customer driven, and that ‘may’ be so – but, I believe vendors have to move away from trying to solve customer problems and concentrate wholly on developing tools that can be selected and used in a generic manner with the ability to be tailored to specific tasks by ‘another’. Put another way; a modeling package that can achieve both history and non history activated by the user as a when appropriate. It may be an oversimplification but the main reason I still prefer to model many things in AutoCAD/MDT is because I can freely and easily (in one place) mix 2D, dumb 3D, parametrics, history and non-history then give native AutoCAD files to any customer using AutoCAD and know he can modify the files as and when he see fit. Sure there are some losses (substantial some times and other times not so) but to me it is just a task that has to be considered and managed and when done with ones eyes open is not as ‘expensive’, or as problematic, as it may first appear.

The larger, more important, forgotten issue is to do with bringing ‘new’, maybe young, workers into the design and draughting world and it’s here I seem to clash heavily with the views of many, existing CADD users, in particular highly experienced 3D users. I believe the starting point, of CADD knowledge and application should ALWAYS be 2D, then moving to 3D and, importantly, knowing when and why to move back and forth between the two when appropriate.

Training - commercially and in institutions - forms a major portion of my work and I see, first hand, the problems created – and largely ignored by vendors and 3D protagonist - by the ‘verticalization’ of CADD and the ‘too’ early introduction and inappropriate use of 3D by (in)experienced design and drafting staff.

We established users; have to start moving vendors in a direction that ensures their products allow a much more flexible approach. As much as that is a bitter pill to swallow, at the end of the day, thorough integration of CADD systems MUST mean all individuals and companies MUST be able to a select a set of 2D and 3D tool(s) that are fully integrated; ones that allow users to move up and down and sidewise at any time. The losses moving in one direction and or the gains of moving in the other should be learned, borne and managed by user(s), not steered by vendors trying to ‘second guess’ what ‘all’ users might/must do based on what ‘one’ other has asked for or, that the vendor wants to sell.

Tough call maybe, but CADD vendors are supplying design and documentation software TOOLS to those who design, and document, every thing CADD developers own and use, except software, and it is high time they started to think as we do. GM does not tell a software developer or me how, to drive their car, or to work, why are software vendors trying to tell users (and me) ‘how to’ design or document a machine or product?

Solid DNA said...

Hi Paul

From your first extract quote,is too early to presume of anything but, after review the videos available, we can see the small plus sing beside some feature. This make me thing they still track 2D geometry. This is where they can tell;

“a two-way street; the ability to move back and forth seamlessly between the two modeling paradigms.”

So how do you “move back and forth seamlessly between the two modeling paradigms”

Again is too soon to presume of anything but from my point of view (Solid Edge user) we talk about Dynamic Edition.

This seem to be the same approach made by SW who made 3D something pass for something similar to ST

You then write:

”When creating 3D geometry, a CAD system is either capturing information about the modeling operation; history-based, or it is not; history-free........"

Has mention in another comment on your blog where i point out lexis on my blog.
And yes is is a question of sementic, but was not after all the point of confusion?

This is why i prefer use the term History each time information is store in a tree. Vertical or Horizontal

Then talk about linear or non-linear modeling. Linear if the order of the modeling orperations influence the resulting part.

So like any 3D modelers ST has history so no need to mention it except if we compare to Autocad like software who has no history.

If we go the long way, ST is a history non-linear modeling or like Siemens call...history-free,
feature-based modeling technology

"...you may go back to the feature tree to modify the parameters" under Inventore fusion. It may be thru base on what i write above. The seem to keep track of the 2D on the back. Loose the 2D and the feature will be corrupt.

Under ST we have move from feature modeling to Procedural feature modeling. Base on Dan Staples answers in other blog, we should see more of them in the next release.

The Line between those two ( base on how i understand it) are very thin but again the big picture that make the difference is the fact that no linear depencies exist. However, few exceptions exist to confirm the rule. Example a pattern using the surface of another SET will have some sort of linear depencie unless it is detach.

Well i am not in the good secrets for any of those soft, but this is how i see things taking shape.

jonbanquer said...

"but, we all have to think more broadly about the wider effect CADD has on the development of our industries and on individuals."

In my opinion this will never happen. Should it happen... yes. Will it happen... no.

"It may be an oversimplification but the main reason I still prefer to model many things in AutoCAD/MDT is because I can freely and easily (in one place) mix 2D, dumb 3D, parametrics, history and non-history then give native AutoCAD files to any customer using AutoCAD and know he can modify the files as and when he see fit."

I understand this completely and I like and agree with your approach. However, there is no way 99.9 percent of SolidWorks users would agree with you based on what I've seen posted over say the last ten years in SolidWorks groups or forums.

"We established users; have to start moving vendors in a direction that ensures their products allow a much more flexible approach."

No doubt about this. Siemens already gets the message and unlike Autodesk they prove it with actual software rather than the vaporware that is Inventor-Fusion. In addition Autodesk still heavily censors their blogs, forums, etc. SolidWorks Corp. has become just as bad.

"why are software vendors trying to tell users (and me) ‘how to’ design or document a machine or product?"

Because for many years CADCAM companies have been able to easily get away with this and they still can.

Jon Banquer
San Diego, CA
http://jonbanquer.wordpress.com/