Direct vs History based modelling

[Anonymous]

There is another thread going with this sort of conversation but that thread is more to do with history based modelling than direct modelling.

I find this topic very interesting and have had this conversation a number of times with no real conclusion.

I would be forever grateful if somebody could give an example of how history based modelling is needed from a mechanical engineering point of view.

Ideally what I would like is a solid example, such as if a person was designing a mechanical device how changing something downstream would not be possible with direct modelling, or some other example please?

[cekuhnen]

@daniel_lyall education and money not a good mix based on my experience.

Coming from a country were education is paid differently for only academic performance was what made you succeed.

This gave schools one dramatic advantage compared to here in the US - without being dependent on tuition you simply

can fail who does not deserve to pass.

Thats an inconvenient truth. I am actually not against something like ITT or community colleges if they would simply be

able to follow academic standards. But truth was that they were more in for the profit than really education.

In my country 70 to 80% of the people receive their training through a thorough 3 year apprenticeship education.

At the end you have to pass the guild exam. If you fail well you fail. If you pass with a B well shame on you.

What frustrates and pains me the most here in the US is how much students have no option besides college as a valid

education and second how much financial costs eat into their study time because they have to work. They already get

less education than in other nations because of the general education classes that have to make up for what is missed

in high-school. I worked because I had the time to do so and wanted the extra design related experience when I was a student.

Anyway - little rant is over. Education is so important - frustrating to see how terribly it is treated.

But hey at least today there is something called internet with an access to an amazing amount of resources not there

in the 90s.

[daniel_lyall]

And people wonder why Germans are so good at making stuff, Education. hell the American education system is one of the worst in the OECD 

[cekuhnen]

@daniel_lyall Many say it is out culture and I think it is over rated.

We can be pretty lazy as well.

To be honest I think what only contributes to that myth is that we simply have a dual education system

that is pretty cut n dry to pass or fail. And the classes to repeat or semesters do not cost anything.

And if you don't get into high-school (hard to get into) you can do an apprenticeship and after 3

years still go to college.

So it offers you a lot of options - you just need to show the academic performance.

However also standards in Germany are falling.

[daniel_lyall]

Money is the root of all evil, 

[jeff_strater]

This is good info, @O.Tan.  No, I will never say never for your vision of advanced direct modeling, but I also cannot promise anything soon.

This is interesting detail on the SolidEdge functionality.  I have not paid much attention to this product since the first Synchronous release.  It sounds like they have continued to make progress.

What you describe is somewhat close to an approach that we discussed early on in the "parametric wars" in our team.  The basic idea was:  All designs were inherently parametric.  However, if you switched to DM mode, we just would not show the timeline.  That way, you could toggle back-and-forth between direct and parametric.  If you started in DM, then decided later you wanted the ability to edit an early sketch, just switch to parametric, and voila:  all the stuff you did in DM was recorded.  However, some of us (me) had lots of concerns about that approach.  When editing a design using DM interactions, you do a lot of stuff that you probably don't want to be recorded.  Each little tweak to the model would then show up as "feature".  It seemed that you would get some ungodly long and incomprehensible timelines that way.  Plus, we had this fear that whenever you lie to the user ("sure, this is a direct modeling design"), you end up getting caught at some point.  There seemed to be no way to avoid introducing "errors" into your design, which have no place in a DM design.  So, eventually we abandoned this idea completely.  But it sounds like SE is somewhat close to that.  It would be interesting to play with that system some time to see if we could get it to misbehave.

Jeff

[Beyondforce]

@jeff_strater,

I have an idea for the unaesthetic side of the Timeline when it gets very long and confusing.

How about Grouping the history in the TL based on the components. Every time you create a component, a similar icon will be shown in the TL, and when you Hover the mouse on top the icon (component) it will expand automatically and contract when you move the mouse away. If you are editing something, then the icon (component) will stay expanded until you finish.

This way, the TL will be better originated, short, clean and not confusing!

Ben.

[cekuhnen]

@jeff_strater

Jeff what you say makes sense. 

To be honest I can only imagine what Phil and there talk about SE since I do not have access to it.

i can see the beauty of the synco]ronos edit - while not knowing if this remains being what we consider DM.

since I come from apps with design histories anyway the timeline is quite welcome to me.

it fits my workflow perfectly. I know the more relationships you set up the more they can become a problem but

i find with work experience you will learn how to use it best and also how you start approaching surfacing as a startagy

BLocking out proportions first and later thinking about details and adding them.

i can see that for someone that is also new the timeline is quite a temptation to put everything in overloading it and making it fragile.

but isn't that the case with all systems that use some sort of an automated tool that base does on relations creates work for the designer?

I have to admit that I do not deal propably with as big parts as maybe phill does.

what is quite elegant in fusion with the how it presents options is that you can also start with the history and then later bake it into a DM design if needed.

[cekuhnen]

@Beyondforce

Even when you use the group function sooner or later it will become confusing to read wit opening closing all those.

The problem is just that a complex design will sooner or later just fill up the timeline.

you can also give names to them.

using components is a pretty good additional way to sort the timeline.

in fusion when UI select a surface it highlights also the feature being used helping you to understand it.

[kb9ydn]

---

@Beyondforce wrote:

@jeff_strater,

I have an idea for the unaesthetic side of the Timeline when it gets very long and confusing.

How about Grouping the history in the TL based on the components. Every time you create a component, a similar icon will be shown in the TL, and when you Hover the mouse on top the icon (component) it will expand automatically and contract when you move the mouse away. If you are editing something, then the icon (component) will stay expanded until you finish.

This way, the TL will be better originated, short, clean and not confusing!

Ben.

---

Ok.  This touches on something that I find to be a major weakness with how the Fusion time line works (at least in my opinion so far).  This is hard to explain and my description of this may not match what is actually happening behind the scenes but I'll continue anyway and hope that I'm not too far off.  

Within a given Fusion design you are only ever working within a single time context, which is the main (highest level) assembly.  How many levels of components and sub-assemblies you have makes no difference; there is still only one working space.  You can "activate" a particular component and do things within that component, but when you reactivate the main component (highest level) everything you did in that component shows up in the time line, AT THE END.  So what you end up with is a linear record of *everything* you have done in the design, in the order it was done in.  This may sound reasonable at first, but it also has some serious consequences.

The first is that even with moderately complex designs the main time line can grow to epic proportions.  I have designs in Solidworks that have hundreds of components (which is not that many to be honest), yet the component tree in the main assembly still fits on one screen.  This same design in Fusion would have a time line that goes on for miles!  Trying to work with a time line that long would be a major hassle.  So what does Solidworks do differently?  Well basically it compartmentalizes the time context of each component or assembly so that the parent assembly doesn't *see* what goes on inside a child component or sub-assembly.  In other words each component or sub-assembly has its own separate time line.  So when the solver recalculates the model, it solves everything in a hierarchical order starting with the lowest level components and then working up the tree until it gets to the top level.  The advantage to working this way is that you never end up with a time line (or assembly component tree) that is too complex to work with because you can always break it down into smaller assemblies (and therefore smaller time contexts).

Another consequence of Fusion's single time context (and this drives me up the wall) is that editing features of a component that were done in the past REQUIRES that the entire time line be rolled back to when those features were created.  Of course this makes sense in order to maintain the single time line consistency.  But I (and I suspect many others) find it a major distraction, really to the level of being counterproductive when working with any design that has more than a handful of components.  I want to be able to edit a component WHERE IT IS NOW, NOT WHERE IT WAS A WEEK AGO.  Now I suppose I could maybe get used to this sometime in the future, but I kind of doubt it.  

Now to be honest, I should mention that Solidworks doesn't keep track of component or sub-assembly positions in assemblies AT ALL; which is part of the reason why Fusion bugs me so much with how it handles assemblies.  From this perspective I find that Fusion's history based modelling implementation is actually much less flexible than Solidworks.  It's so annoying that it makes me want to not use history at all.  But then you loose so much parametric-ness that working with more complex designs becomes very tedious.  This is an area I need to explore more though.

C|

[cekuhnen]

@kb9ydn

I see your point. This is why I rather see Fusion being a generative cad app and not a strict parametric. It can like expressed in our views have problems because how one step follows the next.

for me this is very welcome and logical - it actually also helps me teaching proper design surfacing thinking and strategy planing to my students.

but I can see that sometimes UI want just do well adjust something no matter of the time it was created. But when you know the design well I feel patching the change into the correct timeline position isn't hard.

not saying  you are wrong - but the longer this conversation goes on the more i ask myself why people struggle with the timeline, because as I stasted before I don't know a cad app in my price range that does it all and perfectly.

one has to -take apps for what they are.

currently I am re evaluations SolidThinking to see if we can replace our other surfacing app with it. And SolidThinking has some funny ways how to do things.

yet how surfacing is invisioned in SolidThinking it makes sense meaning what I know so much from alias and rhino I cannot just transfer to SolidThinking.

i feel the same is true with fusion.

when I started with it I asked a lot of why does this not work till I realized that the way how fusion works you simply approach solving that problem differently.

[jeff_strater]

@kb9ydn,

The history-based assembly aspect to Fusion is something that is unique, as compared to other CAD tools.  It does take some getting used to, I admit.  And, there are pluses and minuses to this style of modeling.  On the "minus" side, as you point out, the timeline gets very long for a large design, all in one unit.  And there is the time-warp aspect to editing features/sketches, including component moves.  This is actually both a plus and a minus.  On the minus side, there is the "when I edit this feature, the component moves to a different place", which can be disorienting.  On the plus side, you can do "position-based modeling", where you can position a component, project edges from that component somewhere else, then move it again and project those same edges in the new orientation, and use both of those projections to make new features/components.

You do realize that you can use Fusion to model like you did in Solidworks (the bottom-up workflow you describe), right?  That's why we added the ability to insert designs into other designs.  So, you can localize all the features for a component into one document, then insert them all in a single top-level assembly document that has only assembly operations in it.  Then, you will get a very short timeline for your top-level assembly.  But, IMO, one nice thing that Fusion gets you is you don't have to create separate designs for each component.  You can create lots of components in a single design.  This results in a lot less document-level management to be worried about.  But, that comes at the expense of potentially a long timeline.  Or, you can mix and match:  Some local components, some external.

Obviously, I'm biased here...  Others' opinions?

Jeff

[PhilProcarioJr]

@jeff_strater

I am currently working on a project with Phil.E that is an enormous project. I have listened to all the discussions here about Timeline VS DM and to be honest I still don't understand. I have made modeling Changes to hundreds of pieces on the fly to test different designs in the project with no slow down to my creativity or speed of creation at all and I might add everything is to scale and it is large. I can add there is absolutely no way this project could be done with a timeline and to be honest it would be the biggest nightmare ever if I did use it. There is also a way to make DM act like a parametric model...not 100% but close enough for most needs....took me 7 months to figure that out and no I'm not going to get into that discussion here, just wanted to point that out. Again I am not against the timeline but I still stand by what I said before it's not all it's cracked up to be. I would even go as far as to say that the current implementation of timelines in ALL CAD apps are the worst thing ever added to CAD. 

I would like to add that if anyone out there would like to do a large assembly challenge of Timeline VS DM I will definitely take that challange.

Again this is only my opinion...never said I was right.

[daniel_lyall]

@PhilProcarioJr that just goes back to use what fits.

You have the option to do each part on their own, or have a giant time line that may fail at one point and destroy the model, this happens even on small time lines but can be found easy, giant timeline days to fix maybe.

There is a logical order each type of object needs done, This is super hard to work out, but it is there.

One thing I have seen a lot of people do what I find a why do it that way that's just daft.

Its draw a squire or what ever extruded it up draw some holes on the top extruded them down then draw some more stuff on the top in a new sketch then more extrude then features. (big timeline)

above  3 sketches 3 to 5 extrudes then features

I look at it and go everything could have been done in one sketch and a couple of extrudes, (small timeline)

1 sketch 3 to 5 extrudes then features

useing more than one sketch where 1 sketches could be used there is more chances of a single sketch editing stuffing something up as if you missed one constraint on a sketch on top of the object it could lose its relationship, then its a collapsed across everything.

If it's just one sketch it's harder to stuff up as you can see a mistake.

I work to the rule if the part could be done with more than one sketch what sketches could be on the same sketch then do it that way, it can reduces the timeline by a lot and make it easier to find problems.

same goes for bigger parts with more than one body in there own component what can be use elsewhere and if it gets changed its correct for other parts.

Above is just where the timeline fits and DM does not fit.

DM I use where TL will fail or the part is dead

[TrippyLighting]

---

@kb9ydn wrote:

---

@Beyondforce wrote:

@jeff_strater,

... but when you reactivate the main component (highest level) everything you did in that component shows up in the time line, AT THE END.  So what you end up with is a linear record of *everything* you have done in the design, in the order it was done in.

---

That is not necessarily true!

The control of where something shows up in the timeline is entirely up to you within reasonable and logical limits.

Yes, when you activate a particular component and do nothing other than continue to add things to the timeline then, yes, you'll add it to the end of the timeline.

There are, however, two ways around this:

1. You deactivate the filter that narrows down the view to show only the items in the timeline that belong to the activated component. Then you can place the timeline marker when you want it. That of course is rather cumbersome, but the possibility exists.
2. The better way is to simply pull the timeline marker bak to after the last feature in the activated component and then move it one step forward in the activated component. And voilá you'll be adding the next items to the timeline to the end on that components timeline.

[PhilProcarioJr]

@daniel_lyall

The current project I am working on needed 392 sketches and it is only maybe 10% done. I figure by the time the entire project is done it will have required over 3000 sketches....

You actually just enforced my point, Timelines are only good for simple stuff...granted my idea of simple and many other peoples idea of simple are very different.

I support the use of the timeline for simple parametric stuff.

[daniel_lyall]

1) Design intent.

2) what is it.

3)Should I do it in a DM or a TL design.

This leads onto

3 a) Would I use it in different designs.

3 b) Will I ever change it.

3 c) Will a part of it be used again.

3 d) Do I need parameters.

Then this needs thinking about.

4) can i do it in different files for X ref

5) Should I do it in one file.

This is what I stick to, you guys are the ones that teach most people that's why sometimes with some things I will agree with @cekuhnen Why, stuff that are his strengths @PhilProcarioJr samething @jeff_strater same thing, the other jeff's same thing, Mark, trippy, seth so on.

If I need to know how to do something if you spend time on the forum you learn who to goto for help.  

This is why I agree with you last post @PhilProcarioJr For how fusion works today it is correct. you could do it with different files but that would be a pain (keeping the timeline small)

For the final design 90% of what I do needs a timeline, the other 10% does not matter.

[PhilProcarioJr]

@daniel_lyall

All my die work at the forge I use the time line unless it's T-Spline work, sorry but no matter what anyone says using the time line with T-Splines is absolutely foolish.

My personal work I have no choice...I don't do simple designs and Fusion can't handle the work I do with a timeline....it's just not possible unless A LOT changes with the performance of Fusion.

[daniel_lyall]

Yep fusions strength and weakness, for a noob or a new user This bit is Hard.

In 2 years it will be different again, this comes up every few months 

[Beyondforce]

I disagree with you @daniel_lyall.  Strength and weakness are within the person who uses Fusion. Fusion 360 is a tool and how you use it depends on your creativity and your ability to think out of the box. Moreover, newbies (that includes people with previous experience on a different CAD software) needs to learn the Fusion 360 basics first, which many don't spand the time to do so!

Ben.

[daniel_lyall]

@Beyondforce not realy, yes the person has to learn the fusion way, ounces that is done you stick to fusions strength and weakness.

What is not easy to learn but if you keep a eye on what the EE do and say, it's not that hard.

You cant use DM to do furniture, unless you wont to do a design a few 00 times, user parameters.

Wheelchair add ons built to a person cant use DM. unless you wont to do a design a few 00 times, user parameters.

disability addes built to a person cant use DM. unless you wont to do a design a few 00 times, user parameters.

motor mounts if one off whatever goes if production no DM why standard measurements across different size motors.

if there are standards why redo the wheel.

IF it's a giant design A TM is super hard work, DM is a bit easier.

knowing the above, the first 3 I learnt the other 3 came from another user from recommending what works best and easier in fusion.

you can do everything in DM but somethings are better and easier in TM.

A person has to learn what can and cant be done in the program, I can do a lot in my head but getting a program to do that thing is a, learn how the program will do it in a efficient and a stable way.

IT'S the person learning limits of the program not their limits, this is a problem in some people why they like being in the box not what is around it look up.

being a supervisor and a teacher for a long time you learn human nature.