Welcome to the Onshape forum! Ask questions and join in the discussions about everything Onshape.
First time visiting? Here are some places to start:- Looking for a certain topic? Check out the categories filter or use Search (upper right).
- Need support? Ask a question to our Community Support category.
- Please submit support tickets for bugs but you can request improvements in the Product Feedback category.
- 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.
Comments
- Direct editing is very helpful for manipulating imported geometry (no features present).
- Direct edits can come in handy for making changes 'after the fact' in a parametric model. E.g. You modeled ribs in a plastic part to have 1mm thickness, then the molder suggests a thickness of 0.85mm. You do the change with a direct edit appended to the feature tree. (I have found this to be a great way to keep track of design canges as well, regardless of any PDM system used.)
- It is impossible to omit paramtrics entirely! Whenever configurations are involved or intelligence in how design features are related (e.g. "This hole should ALWAYS be 50mm from that hole."), parametrics are essential. You will always hit a wall with a pure direct modeling solution in a real production environment.
Just my 2cts.Dries
"Well defined" Sketches to do solids to do layouts??
I know CAD programs that can do very complex plastic parts without any parametric trace.
For some designs, parametrics are best. For others, a pure direct approach is best. Direct modeling is indeed very welcome in (some) initial conceptual design!
However, you can't cover all design problems with pure direct modeling. How do you handle configurations/design variations in a direct modeler?
All pure direct modeling packages currently in the market are niche players and have been for many years.
I'm interested in knowing your view.
Dries
Parametrics aren't going away anytime soon.
However, I would certainly appreciate expanded direct editing tools in Onshape.
Dries
direct modelling...Spaceclaim..fantastic app for certain tasks. But Rhino is a direct modeller as well, for the most, but really, so tedious for doing complex tweaks to form on a developed model.
As Peter says, history based modellers like SolidWorks wipe the floor with pure direct modellers for logic based design (aka Driveworks type use). Seriously. I did a benchmark a couple of years ago for a customer where we took the SolidWorks files of a product which was driven by 2 sketches and linked patterns, with linked drawings to that. We tested Spaceclaim, SolidEdge and Inventor.
The task was produce a new design in under 30 mins with linked approval drawings.
Spaceclaim? All that selection, moving etc etc. Took 2.5 hrs.
SolidEdge...took 40 mins...using History. 2hrs using ST.
Inventor...took 30 mins
SolidWorks...25 mins.
but its not just for logic based design tasks. Try driving a complex surface model using direct only. Its not like we don't try either. We use Rhino, Tsplines, Modo alongside history based systems like SolidWorks and Shark. Sub D is not the answer...only applies to a small proportion of design work. Production level surfacing is about subtle control, accuracy, editability. Some of that you can do with direct, but only with VERY costly direct systems!
So much of the "the future is direct" bandwagon is based on editing prismatic objects. Any decent CAD system can do direct operations. I see so much toss at demos importing dumb solids and saying "look, we can edit this fillet". Wow. The $99 ViaCAD can do this. Nothing new.
So, I'm not wanting to get into a big argument on direct vs history. It is irrelevant. We need both approaches in one product. Not one or the other. Both.
Kevin Quigley said:
- "...Any decent CAD system can do direct operations. I see so much toss at demos importing dumb solids and saying "look, we can edit this fillet". Wow. The $99 ViaCAD can do this. Nothing new..."
Direct Design is not just modify, Is mixing, complementing, interacting with wireframe, surface and solid ALL in one enviroment.www.youtube.com/watch?v=onXOVXkV-Zs&ab_channel=ManuelValenzuela
I also think you misunderstand what I mean by iterative production surfacing. That video of the shoe sole is a great example of what I call static surfacing..you create the curves, create the surfaces, the to edit it you basically rebuild the areas you want to edit. You can take that approach on any CAD system worth using. This was exactly the approach we used in the late 80s. Not new. This is essentially the same approach Rhino uses...where if you are lucky you can maintain associativity for a feature. This does have its place but for iterative design it is a tedious innefficient approach.
what I mean by iterative design is creating the curves, creating the surfaces/solids then iteratively editing the shape (not just fillets as in the helmet example), the whole or localised areas of the shape, but tweaking the driving curves or directly editing the surface CVs. If you want examples of a similar to KeyCreator product but one that has full associativity from curve to surface to solid look at SolidThinking or SharkFX.
But the bottom line is this. We use Solidworks, Modo, Rhino etc because each has its own benefit for different design tasks. People who say one solution and one approach meets all needs are either selling a product or working in a limited methodology in a restricted design area. We don't. We design everything from bathtubs to speculums. One solution does not fit all.
Finally, you can in fact drive a surface with dimensions, very easily. SolidWorks has the cabability to drive CVs on splines with dimensions, as does many other systems. And very useful it is too.
So lets not get into an argument over this. It is only CAD after all, and most people here have been doing it a long time, and seen most of what is on the market. What we want is a new approach, not a rehash of old ways.
If OS would choose to go just direct modeling from the beginning and wouldn't develop perfect sketching environment - it would be too different to easily learn for people usign mainstream 3d modelers and propably would end up being just another weirdo on the web.
At first we need true cloud based alternative to current market leaders - then OS can show new directions for design workflow too.
To design new things, it's easier un direct modeling (just take a look at Ptc Creo Elements direct modeling or Ironcad). There are no parameters or contraint in sketch. 3d parts are stretched like un 2d cad... If you have to built tour machine more than once with small change parametric is the way to go
But parametric cad softwares are better and faster for machine production. In the past I worked for a company which was making "900 series" conveyor... These " 900 series " conveyor were always built the same way except fort width and length... In parametric cad, 2 parameters need to be changed and the modeling and drafting work was done... You have to think which paramètres are required but when it's done it's very fast to generate drawing...
It depend on what you are designing and will you design it on a regular base...
Sorry for my mistake (my primary language is french)...
I think this discussion is one where individual users are seeing something in the OnS business model which appeals to them, and thinking that their particular area of design is (or should be) the target market.
Specialised packages will always outperform more versatile ones in their area of specialisation.
I personally get involved in all the areas mentioned above, and dozens more, and the compromises involved in using a parametric modeller for ID pale into insignificance alongside the difficulties of trying to use a modeller optimised for ID to design engineering componentry and mechanisms.
-Engineering works in a parametric world
-Then you branch, turn off parametrics and release to manufacturing
-Manufacturing can tug and pull the model into place using direct modeling
-Then you merge the direct modeling changes back to engineering (parametric world) for the follow on r2 design.
I find it crazy what manufacturing does to my engineering layouts. They destroy it every time. Most companies don't archive the engineering layouts and only control manufacturing documentation. Finding the latest engineering layout is always a pain especially if theres 400 pack-n-go folders and you can only sort based on modified date. The existing solution industry uses is half-baked.
To defend manufacturing, they have no idea what the design intent was or how the parametrics were put together. It's really impossible for them to work on anyone's layouts and therefore they strip out all the goodness. Once it's gone it never comes back and the dance steps repeat over and over again.
I realize there are guys like Bruce that have figured out a tight system and can control everything for a small production shop. But for most companies, it's not working.
What's the craziest thing manufacturing has done to your engineering layouts?
-remove all assy references
-save everything to a parasolids file
-circle interpolate splines
-un-virtualize an assembly
I'd like to see one model make it through engineering->manufacturing->engineering........
I haven't seen this work, ever.
I hope direct and parametric can be married together in Onshape... intelligently. Could be awesome. I am becoming convinced that feature based history methods are not nearly as superior as something else currently out there. I wish, I wish.
Hi Everybody, I just read you and I would like to add my personal feedback about direct modeling technology.
I'm working on SolidEdge and mostly designing electronics parts.
In my company, references are organized with families of product (40 pieces in 1 family). Each family’s assembly has mostly the same design, only 1 or 2 dimensions are changing from one to another.
To model them, I'm using the parametric technology. Indeed the fastest way to build all models is to create an adjustable master model (native format) and extract the slaves (for release: Step, Iges...) Time => 15-20mn for basic parts from 0 to 40 assemblies done and extracted.
=> I've try to build an adjustable part with direct modeling technology and many problems occurs:
-You lose the link between sketches and functions.
-Pattern parameters are limited.
-I need to add dimensions on symmetries. If I don't, it moves when I adjust the part. (move it manualy)
-Till I want to make a dimension adjustable, SE switches automatically to the parametric mode and shows an error message.
Basically I can build exactly the same part with synchronous technology, but it will take 20x more time because I have no way to make my models adjustable automatically. So I must draw them one by one. It makes this solution not viable for my job.
=> For the kind of design I would like to do, the following parameters should be apply: (starting from direct modeling mode)
-Keep a link between sketches and functions
-Pattern features should have the same parameters available as in parametric mode.
-Adjustable dimension must be editable in direct modeling mode.
Good points for DM technology => Drafting liberty and editing facilities.
Bad points => Geometry lack of control leads to an impossibility to create adjustable parts.
A mix between both technologies might be the best, but here we have a turning point between users. Indeed we do not all expect the same behavior. The solution might be to entirely set up the software behavior for each use.
The issue is the barrier between direct modeling and parametric mode, maybe we should search ways to make them work together instead of saying: "This one is the best!"
Parametric design is about incorporating design intent, particularly (for the purposes of this discussion) with respect to likely future changes or variants.
So ... in cases where the future of a model cannot be predicted, or where variants will not be generated, the extra burden of full parametrics may not be worth carrying?
I do the link between my current activity and the title "forget parametric".
In the beginning we use the direct mode to validate the design => Variants not predictable => too many constraints in parametric mode as you said.
Then the parametric one for "slave’s creation" => Variants is knew and I just need several models with different sizes => not enough geometry control to set up "automatic" models
So, in order to reach the idea: "forget parametric" I suggest it might be a good thing to have it in one mode which allowed both best features.
Which would means to be able to set up your software behavior to be compatible to everyone's needs.
I do think there could be some awesome things in this area, though.
For the topic:
I have tried direct modeling few times but I don't see a reason to 'Move a face' since I can just set new value for parameter to achieve the same.
Only thing it has been good is modifying imported files without history.
Don't get me wrong, I'm not against direct modeling - I just don't know where or how to use it effectively. How do you begin from blank screen with direct modeling? How do you create anything precise without dimensions and if there is dimensions what's the difference then (dimensions are parameters, right?)?