Welcome to the Onshape forum! Ask questions and join in the discussions about everything Onshape.

First time visiting? Here are some places to start:

  1. Looking for a certain topic? Check out the categories filter or use Search (upper right).
  2. Need support? Ask a question to our Community Support category.
  3. Please submit support tickets for bugs but you can request improvements in the Product Feedback category.
  4. Be respectful, on topic and if you see a problem, Flag it.

If you would like to contact our Community Manager personally, feel free to send a private message or an email.

Forget the Parametry

13»

Comments

  • jon_30jon_30 Member Posts: 37
    edited April 2015
    "I do wish that Onshape had started DM because I'm more of a DM fan and if their effort was put into DM from the beginning I can only imagine how excellent it would be moving forward!"

    What greatly disturbs me is I see more original ideas for Direct Modeling in other products. Specifically, what use to be called Solidworks Mechanical Conceptual and IronCAD. Here is a video that does a great job of illustrating exactly what I mean. I like how the direct edit is done locally and not just tacked on to the end of the history tree:

    https://www.youtube.com/watch?v=ANtg4Gpc310

    Jon Banquer
    CADCAM Technology Leaders group on LinkedIn


     

  • kiwileekiwilee Member Posts: 8
    I think there is a huge difference in how each company allows its staff to use CAD.
    And where the design office is largely affects its core role.
    You can be in an office designing factories or steel framing largely removed from the machine shop. 
    Or within arms reach of the CNC, cutters and welders.

    Two years ago I was tasked with setting up a design office in our company. (60 staff total)
    We have no CNC or lathes and any drawings were previously on Publisher. Yes... Publisher.

    I was to be producing steel assembly and fabrication designs + Roto moulded products and moulds.
    Lots of variation.

    I trialled Inventor for one of our projects a multi hull Kayak/Catamaran.
    The parametric design experience was something to behold. And in the end stifling so
    we purchased SolidEdge ST6 and could then operate in either Parametric or Synchronous at any time on any model part.

    When I was on the (1 day) training course I asked the guys at SE if I should primarily use the ST mode or stay in PAR.
    They advised as we had no heritage CAD history and our office was starting fresh, just go for it in ST.
    I have to say it's been brilliant. 
    Parametric has it's place but ST or what you are calling Direct modelling is brilliant.
    About the only time I use Parametric is when one of the plugins requires it. Eg when adding working Gears from a gear generator plugin.
    Being able to have mix and match of Parametric and Synchronous Technology within designs is excellent.

    I don't dream of converting  a highly disciplined parametric CAD user over to be 90% Direct overnight.
    Or expect machine workshop integrated workflows to change quickly either.

    Having a well constrained database driven directly editable model is a reality already. It is how I work now.
    It is not like using the paint brush in Photoshop.
    It is still dimensioned and constrainable CAD.


  • alin_1alin_1 Member Posts: 8
    edited April 2015
    This is an amazing video, Jon! :)
    " Here is a video that does a great job of illustrating exactly what I mean. I like how the direct edit is done locally and not just tacked on to the end of the history tree:

    https://www.youtube.com/watch?v=ANtg4Gpc310

    Jon Banquer
    CADCAM Technology Leaders group on LinkedIn


     


  • rossross Member Posts: 8 PRO
    lowjack said:
    @Onshaper - I don't have any real criticism of direct modelling one way or the other, I've spent precious little time working in a direct modelling environment, so I am hardly qualified to offer an opinion on how great it is. My criticism here was the topic itself - Onshape is advertised and sold as a parametric modeler, so why does anyone use it and expect it to be a direct modeler? I don't buy coffee and complain because it doesn't taste like beer, or buy a pickup truck and wish it did a better job as a station wagon.  

    If Onshape can build direct modeling into the package successfully, great. I'd look forward to getting my feet wet in that area.


    When all the other players either have or are introducing direct modelling I am at a loss to understand why Onshape have only done the minimum necessary to be able to edit imported geometry. With all the potential of Onshape to overcome collaboration issues I find the lack of DM perplexing to say the least.

    Would appreciate some serious comment from the Onshape guys.

  • 3dcad3dcad Member, OS Professional, Mentor Posts: 2,377 PRO
    edited April 2015
    @KiwiLee Can you check my questions in previous page. I would appreciate if you would have time to answer as you have probably bumped into situation where you would need similar functions at your work. I'm still struggling to see the benefits of DM over parametric.

    Looked at some youtube videos last night and bumped into this, don't know how it's related to DM but this is what I would like to see in Onshape assembly:



    It would be really nice to use cad for testing how things work, not just trying to replicate something on screen that you already know 'in real world'.
    //rami
  • Charles_22Charles_22 Member Posts: 2
    Just my two penn'orth:

    Parametric is everything when months, even years, later a mod deep in the design ripples outward.

    One of the great banes of CAD has been the attitude that because everybody else does it then so should we. This has lead to some amazing bloat in some mainstream products.

    Keep It Simple Smart.
  • pete_yodispete_yodis OS Professional, Mentor Posts: 666 ✭✭✭
    ross said:
    lowjack said:
    @Onshaper - I don't have any real criticism of direct modelling one way or the other, I've spent precious little time working in a direct modelling environment, so I am hardly qualified to offer an opinion on how great it is. My criticism here was the topic itself - Onshape is advertised and sold as a parametric modeler, so why does anyone use it and expect it to be a direct modeler? I don't buy coffee and complain because it doesn't taste like beer, or buy a pickup truck and wish it did a better job as a station wagon.  

    If Onshape can build direct modeling into the package successfully, great. I'd look forward to getting my feet wet in that area.


    When all the other players either have or are introducing direct modelling I am at a loss to understand why Onshape have only done the minimum necessary to be able to edit imported geometry. With all the potential of Onshape to overcome collaboration issues I find the lack of DM perplexing to say the least.

    Would appreciate some serious comment from the Onshape guys.

    @Ross, it's still early days with Onshape.  No other CAD platform is Full Cloud and non-file based.  This has overwhelming advantages on the data management/collaboration side of things.  Onshape did a great job of nailing that first and foremost.  They are developing a CAD package around that platform and NOT bolting a PDM system onto to a file based system.  This will allow the CAD advancements, I believe, to come quicker over time.  Drawings is close.  Having someone provide huge amounts of the functionality out of the box should again allow Onshape to focus on things like geometry if we users are going to push them to.  Keep pushing.  They are aware.

    There are lots of terms flying around this thread, but for what it's worth, I think of Direct Edit or DM as free-form pushing and pulling of surfaces without any constraints involved.  History based, to me, is what people would refer to as parametric.  Synchronous Tech is unique at the moment and is really neither of these as I understand it.  @KiwiLee has mentioned it is still constrainable CAD.  @Charles is right.  Keep it as simple as it should be.  We don't want Onshape to rush into something because we are a little too impatient - I don't think they would either.  Onshape is filling out the History based methods at the moment.  The direct edit tools are helpful as they are now.  Something like Synchronous Tech, if it's worthwhile, would be another tool in the arsenal.
  • onshaperonshaper Member, Mentor Posts: 98 ✭✭✭
    edited April 2015
    Just out of curiousity, in DM:
    - Can you do pattern/mirror and come back to edit in same dialog?
    - Can you set relations to dimensions ie. D3 = D1+D2?
    - Can you have global parameters or spreadsheet driven designs?
    - Can you have configurations (where difference typically would be different parameters and having some features suppressed)?
    - Can you have named dimensions and show them in BOM?


    Yes
    Yes
    Yes
    Yes
    Yes

    Note: I can only comment for SolidEdge Synchronous.

    Here's another thing; lets say you're doing dimension optimization based on FEA results. In parametric if you are optimizing an early feature it must regenerate the entire model/assembly to change the dimension. In DM, the edit is far quicker. Sometimes on the order of 5-10 seconds quicker. So for parts with longer history trees the difference in time for a dimension optimization can be 30s in DM and 600s in PM.
  • onshaperonshaper Member, Mentor Posts: 98 ✭✭✭
    edited April 2015
    Charles said:
    Just my two penn'orth:

    Parametric is everything when months, even years, later a mod deep in the design ripples outward.

    One of the great banes of CAD has been the attitude that because everybody else does it then so should we. This has lead to some amazing bloat in some mainstream products.

    Keep It Simple Smart.
    Hi Charles,

    I was hoping you could expand on this. I have found that with Parametric models, you better have your relationships 'programmed' just right, otherwise an early change has a big chance of blowing the whole thing up on recompute.

    In DM, you can change an early feature without changing anything else. That's the power of it. So in your example, DM is actually a better tool in my experience.
  • pete_yodispete_yodis OS Professional, Mentor Posts: 666 ✭✭✭
    edited April 2015
    3dcad said:
    @KiwiLee Can you check my questions in previous page. I would appreciate if you would have time to answer as you have probably bumped into situation where you would need similar functions at your work. I'm still struggling to see the benefits of DM over parametric.

    Looked at some youtube videos last night and bumped into this, don't know how it's related to DM but this is what I would like to see in Onshape assembly:



    It would be really nice to use cad for testing how things work, not just trying to replicate something on screen that you already know 'in real world'.
    @3dcad That is not DM.  That is known as other things.. physical dynamics in SolidWorks speak.  Some might call it part of motion analysis.  This is doable - and even more computationally supported sitting atop a massive cloud compute platform like Onshape is on.
  • 3dcad3dcad Member, OS Professional, Mentor Posts: 2,377 PRO
    @Pete_Yodis Yep, not much modeling in that but looked really nice.. Is there any fundamental differences in assembly mode when working with direct modeler?

    //rami
  • onshaperonshaper Member, Mentor Posts: 98 ✭✭✭
    edited April 2015
    I find there is more difference program to program than comparing DM to PM.
  • onshaperonshaper Member, Mentor Posts: 98 ✭✭✭
    edited April 2015
    I would describe the difference between DM and PM as this:

    PM: You program a recipe that creates a part. If you need to make a change, you throw your model in reverse, make the change, and throw it into drive and pray.

    DM: You make a part. If you need to make a change, you change the part.


    That said, there are certain things that PM does waaay better like sweeps. You can make sweeps in DM, but it often just turns into a surface that is difficult to change. This is more a limitation of software development, not a limitation of DM.
  • andrew_troupandrew_troup Member, Mentor Posts: 1,629 ✭✭✭✭
    edited April 2015
    3dcad said:
    <SNIP>

    Looked at some youtube videos last night and bumped into this, don't know how it's related to DM but this is what I would like to see in Onshape assembly:



    It would be really nice to use cad for testing how things work, not just trying to replicate something on screen that you already know 'in real world'.
    That's bogglingly well behaved, considering the amount of solving which is having to happen in real time. If that's in our future, bring it on!

    It's one thing to take something which is already well understood, like chain, and create rules to make it behave as we know it should. It's not trivial, and it's rather useful.

    But it's quite another thing to be able to invent a new jointed assembly, based on (say) a hunch, and to be able to preview how the parts would interact without having to make a (real world) study model. *

    So I disagree with the person posting the video, who qualified the physics he demonstrated in Crea as perhaps being limited in usefulness.

    * Of course, 3dcad already said the same thing, just more concisely ...
  • joe_dunnejoe_dunne Onshape Employees, Developers Posts: 144
    edited May 2015
    Well this is a blast from the past. Kind of seems the market spoke on this topic long ago...   History based modeling vs non-history based modeling, Direct modeling is more of a marketing term. So I think it best left out.

    Nobody seems to disagree that constraints are a good thing.  Geometric constraints are good. equations are good. You want behaviors, you want dimension designs etc... Right now Onshape uses constraints in a history based approach.

    Sure history based modeling is good for family of parts and similar but different designs and many other benefits already stated in this thread.  But is more basic than that. 

    I think the fundamental power of history based design is that it breaks down a very complex problem into smaller  manageable chunks.  Anyone who has ever used a non-history based modeler is familiar with one message. 'failed to calculate".   When you don't have a history you are asking for the system to do all kinds of things that border on impossible. Non-history based systems simply fail a lot.  Just look a basic 3D model with fillets. For a history based system its a simple problem, but for a non history based system its a very complex problem, even for a simple shape.

    Also to use a non-history based systems means to throw away sketch driven design... Since by definition having a sketch that drives a feature is called "history".   And the ability to tweak a sketch and have the model update is great for conceptual design. being able to make many many changes in predictable, iterative manor is what history based design rocks at. 

    Non-history based systems simply put all the burden on the human to do all the work. As Kevin clearly showed in his stats. It takes more work to build things in Non-history based tools, because you are doing all the work all the selections all the transforms everything.  A lot more selections. a lot more manual interactions... There is no design intent... 

    You cannot just magically make a imported 3D part smart. Trust me.. I have seen a lot of attempts. (several I personally worked on and shown in videos on this thread).  Yes you can add constraints to a non-history based part.. The challenge is what constraints do you add?  All you have to do is look at auto dimension tools in 2D sketches. Been around for a long time. And the problem is there are just too many different possible solutions... And that in single 2D sketch... how hard to you think it is to do on a entire 3D model? the analogy that is fairly accurate is "Nailing jello to a tree" 

    So the act of adding constraints to a 3D model is a manual process.  Handled very well using tools like move face. These are great for manipulating in a more "direct" way....  You will see more cool stuff in Onshape as time goes on. But I think the 20 year old discussion of history vs non-history has been already been decided by the market...  

    The market has chosen history based modeling.

    Joe
    Joe Dunne / Onshape, Inc.
  • dave_cdave_c Onshape Employees Posts: 32
    Just to add a little to Joe's comments, I don't think arguing about DM vs. PM is particularly constructive.  I think the industry as a whole has not done users a service by turning it into a religious war over the last 15 years.  At Onshape our goal is to provide you a set of tools that let you create and modify your product designs as efficiently as possible.   In our minds, the issue is not whether they have a history or not - the issue is do the tools let you model what you want to model effectively.   Onshape already has powerful direct editing features - you can learn about them in the Direct Editing video and we will add more over time.  Onshape users use them all the time.   Direct editing is in our view not about history - it is about providing tools that let you make the change you want to make when you want to make it.  From that point of view, extrude, revolve, Boolean, draft and many other "traditional" features are also very effective direct editing features.  And this is where the philosophical arguments of one vs the other break down.  Onshape we disagree with the "parametric purists" as well - that say if you add features that change your design intent you are doing it wrong.   Our goal is to give the users the power and the tools to use the right tool at the right time.  They are the best people to decide whether it makes sense at a moment in time to do a new operation to change the model, change a dimension to edit the model, or edit the history to change the model - not us.  Onshape's goal is to give them the tools they need. 

    That said, and as Joe said, there is no denying that using a feature list (or history) is a very powerful and very well proven way to define and modify manufacturing designs.  

    The practical input of Onshape users in the forum has been very helpful to us in prioritizing future work.  And the way the community has been helping newcomers has been awesome.  Thanks! 
  • jon_30jon_30 Member Posts: 37
    edited May 2015
    "Just look a basic 3D model with fillets. For a history based system its a simple problem, but for a non history based system its a very complex problem, even for a simple shape."

    Not a problem with a CAD For CAM modeler like Autodesk/Delcam PowerSHAPE which will recognize the fillets created with direct modeling and create Features for them. It does this with Feature Recognition. It's very fast and it works very well:

    https://www.youtube.com/watch?v=0SI_iSs1qvQ

    If Onshape employed someone with an extensive machining job shop background I doubt I would have to explain this like I have above.

    "The market has chosen history based modeling."

    The market is changing. History based modeling sucks for working with non-native solids.

    After well over two decades we are still dependent on a tree for a history based CAD modeler user interface. I strongly believe that an innovative, new CAD modeler will find a better user interface solution than just using a history tree.

    Jon Banquer
    CADCAM Technology Leaders group on LinkedIn




  • joe_dunnejoe_dunne Onshape Employees, Developers Posts: 144
    edited May 2015
    Feature recognition, direct editing tools, modifying fillets on imported parts...  We agree, hence why we have been doing this in Onshape since day one:

    http://youtu.be/amI0JZRso-c

    Joe
    Joe Dunne / Onshape, Inc.
  • jon_30jon_30 Member Posts: 37
    Your video link is private so I can't view it.

    Jon Banquer
    CADCAM Technology Leaders group on LinkedIn



  • 3dcad3dcad Member, OS Professional, Mentor Posts: 2,377 PRO
    edited May 2015
    I don't know anything about coding cad software or how files different between cad systems but shouldn't the feature tree be saved together with the model in universal format. Then every cad-house could reference their own features to universal and build the tree accordingly?

    I'm sure it's not that easy but I agree with @Jon Banquer that even though history based modeling might be better way than non-history, things should still develop towards full support for non-native models. But currently opening a model without history needs basically almost same work as starting from scratch to convert into native format with history.

    I also agree with @Joe Dunne "power of history based design is that it breaks down a very complex problem into smaller  manageable chunks". It makes things easier when you can concentrate to one little thing at a time and knowing you can change parameters afterwards to adjust if needed.

    And very much agree with @Dave C that arguing about DM vs. PM isn't getting us anywhere. This is the only thread with negative tension in comments and a lot of down votes in comments which is pretty rare in other conversations.

    Everybody would probably just like to have better, simpler, flexible and responsive 3d modeler to handle different type of work to be done effectively. Onshape development has been so fast, yet still well planned with robust features (very few bugs) that I'm confirmed they are able to brings us something better than any of existing cad systems.

    Edit: And I appreciate very much the fact that great number of Onshape people are here with us in the forum discussing about the development.
    //rami
  • onshaperonshaper Member, Mentor Posts: 98 ✭✭✭
    edited May 2015
    I'll agree that if you have to pick one to focus on, you best pick a history-based parametric modeling system. It does everything quite well.

    I will disagree that the market has decided on parametric. Aside from SE:Synch I haven't seen a DM tool worth using and it's changing drastically every year. There is a lot of development left here.

    Having a 3d part with 3d dimensions that behaves appropriately is quite addictive. I've never seen this in PM! Although, I don't see why you couldn't do it.
  • mevalemevale Member Posts: 22
    edited May 2015
    ...Design intent is a term used to describe how the model should be created and how it should behave when it is changed.
    With a parametric modeller it is very important to plan out the design BEFORE modelling..." DS.

    Parametric Design Intent.
    It is... "strange", choose to waste time defining the part requeriments and preview behaves at the very begining.
    (THINKING) ----Drawing--------
    How easy is to change if you go wrong when part is almost finished?
    How easy is to modify a design from a colleague?
    How easy is to modify a parametric part from other CAD systems?

    Direct Design Modeling.
    It is more naturall define smalls requeriments when they appears.
    (t) --Dra-- (t) --wing-- (t)
  • onshaperonshaper Member, Mentor Posts: 98 ✭✭✭
    edited May 2015
    @mevale makes a really good point. I started out DM and when I switched to a PM modeler I was so angered by how you need to know what your design is before you start drawing. Without knowing that, you're screwed. Really screwed.

    DM is like building with Legos. PM is like baking a cake.

    Sometimes it's best to have a plan, sometimes you're trying to use CAD to create a plan. PM and DM both only work really well for one of these. They have their uses.
  • kiwileekiwilee Member Posts: 8
    3dcad said:
    Just out of curiousity, in DM:
    - Can you do pattern/mirror and come back to edit in same dialog?
    - Can you set relations to dimensions ie. D3 = D1+D2?
    - Can you have global parameters or spreadsheet driven designs?
    - Can you have configurations (where difference typically would be different parameters and having some features suppressed)?
    - Can you have named dimensions and show them in BOM?
    Yes to those, that is what makes SolidEdge so useful for me at work.
    I built full cut lists off the BOM. Loved how the dimension data can be manipulated. Named etc. And drives the part and BOM.
    Im not massively experienced with the pattern / mirror but I have used it in large Trailer design where components were mirrored.
    But as those were then unique parts I needed unique SAP numbers anyway so took the easy save out and rename within the assembly.
    Our shop floor has no CNC's so all parts need drawings too. No problem and it's all live and relative to the parts.
    It's working like this that got me excited about CAD again.
  • mevalemevale Member Posts: 22
    edited May 2015
    3dcad said:
    • "...Can you do pattern/mirror and come back to edit in same dialog?..."
    Pattern/Mirror - Yes
    Edit in same - No
    Edit each mirrored side separattely - Yes (Paramatric can´t do that)
    • "...Can you set relations to dimensions ie. D3 = D1+D2?..."
    relations - Yes
    • "...Can you have global parameters or spreadsheet driven designs?..."
    Parameters - Yes
    • "...Can you have configurations (where difference typically would be different parameters and having some features suppressed)?..."
    Feature suppresed - Yes
    • "...Can you have named dimensions and show them in BOM?..."
    Name dimensions - ????

    Those Yes was for KeyCreator.
  • rossross Member Posts: 8 PRO

    Thanks Joe Dunne & Dave C for eventually clarifying Onshape's position on 'Direct' modelling. For the record I don't happen to agree with much of what you said.

    In response to 3dcad's comment 'This is the only thread with negative tension in comments'. May have something to do with the importance of this issue to a lot of people.

  • quentin_4quentin_4 Member Posts: 7
    Hey @Onshaper ,

    Do you have any news about my small challenge ?
    Because you keep saying it is possible to build part's families with ST but I'm still waiting ;)

    Maybe you didn't found solutions ? ^^

    ++
  • onshaperonshaper Member, Mentor Posts: 98 ✭✭✭
    Hey Quentin, sorry I haven't gotten back to you. Finals week here and I'm scrambling to finish up!
Sign In or Register to comment.