12 Replies Latest reply on Apr 28, 2011 6:12 PM by eric fairbanks

    Are SolidWorks surfaces good enough?

    Kevin Quigley

      That might be an odd title for a thread but let me explain. I am currently looking at ways we can speed up and enhance the quality of our modelling work from conceptual design through to production surfaces. As part of this process I have been looking at several different packages and last week I spent a day with an Alias expert to see what that app could offer.

       

      Now I am familiar with surfacing only systems and the general methodologies of working with surfaces - we also use Ashlar-vellum Cobalt, Shark FX and VX here - all of which have good 3D surfacing capabilities - but I am trying to focus everything into Solidworks as I really see that as being the most efficient way. I am comfortable with Solidworks and the surfacing in that but I do find it a bit cumbersome for conceptual design, and frequently wish the spline creation and controls were a bit more reliable and flexible. We have been looking at Rhino as well, but the lack of associativity to surfaces and constraints to curves is limiting. Hence why I looked at Alias.

       

      Any product designer out there will have heard of Alias, and it is still widely used in many big name design businesses. The guy who I visited had worked as an auto designer for many years and then moved into training where he trains all the big name auto companies in Europe - he knows his stuff. He showed me how he would approach a product I designed in Solidworks, and he worked through the process in Alias. We looked at the imported Solidworks surface data and that is partly my question.

       

      I was quite frankly horrified by the surfaces as imported into Alias. To create this particular object I used boundary surfaces and fills - all generated from a two point spline (so literally a point at the start and a point at the end, with directional and constraint controls) In SolidWorks the curves are smooth, the surfaces looked smooth and G2 continuous. In Alias they looked smooth on the shaded preview but the wires showed hundreds of UVs on the fill surface. We then remodelled the worst section of the product in Alias (his comment was - modelling in Alias is slower than Solidworks, but better in terms of quality). He spent some time setting up the curves, ensuring continuity then surfacing. The IGES file I exported from SolidWorks was 11MB. The IGES file from Alias was 300kb.

       

      I came out of the day having learned a lot and thinking that maybe I did need to "up my game" and get into Alias. But back to the thread title - does it really matter? Working in surfaces in SolidWorks you are kind of led a lot of the time - which is good, but you cannot really tell how good your surfaces actually are from a density of points perspective. Similarly, creating 3D curves in SolidWorks is relatively easy, but the controls are nowhere near as intuitive or flexible as they are in say Rhino. The point is I am careful, I take time to set up curves and surfaces. It all looks OK in Solidworks, but is it really? Am I really just seeing the glossy paint covering up the dodgy brickwork? I don't know. All I know is what I saw with my own eyes.

       

      Personally I think if you are designing small objects then it is not that critical. Surface continuity on a handheld part is less critical on something the size of a car body. But on those larger parts (that I do work on) I am starting to doubt what I am producing - and that is a worry to me. Up to now I have always said, if it looks good in Solidworks it is OK - maybe that is not actually the case?

       

      As this evaluation process continues I am seeing some serious defficiencies in SolidWorks that really ought to be addressed:

       

      1. Lack of UV/iso curve visibility - why can't we defined the number of UV/iso curves as a view option? Even the $99 ViaCad has this - every other CAD system I have has this. SolidWorks doesn't outside of surface previews during creation. This really is the first item in looking at a surface.

       

      2. We need more and better evaluation tools for 3D curves and surfaces. Alias, for example, has a great surface shader allowing you to move the lights around, and another that creates an iso curve on the surface at any angle with curvature spline and numerical deviation details.

       

      3. We need better 3D curve creation and editing tools. I use a 2 point spline most of the time because anything else is just too hard to do. Surface demos usually have long slender parts with multiple spline points. In reality many parts are more subtle than that, and require tweaking of points. Again taking Alias as an example they have a lovely scaled move CV command - where you adjust the scale of the effect of the movement of your mouse to the movement of control point - makes for very easy precision control.

       

      4. Display of surface internal points - this was the shocker! Just showing all the points showed up the density of points used by the fill surface and to some extent to the boundary surface. Again in Alias (and others) you can show surface control points as you shape it, adjusting the points to make the "flow" better.

       

      Bottom line. If all this was added to SW premium it would be a no brainer upgrade. As it is I am left thinking perhaps I need to look elsewhere to do this kind of quality surfacing work. Which is not what I want to do, but perhaps what I need to do to get to the next level?

       

      Be interested to hear from others.

        • Re: Are SolidWorks surfaces good enough?
          Neil Larsen

          I agree SW has some shortcomings but really I wouldnt be expecting much/any more functionality to be added to existing SW.

          When I've asked SW in the past about how good things actually are they have always declined to answer.

          Nothing new about that attitude of course, its a little much for some egos to have the quality of their 'product' examined I think.

          Passing suggestions to improve this and that and extend functionality have disappeared into the ether..perhaps to condense out of a cloud some time as ER 4951..

          The best answer I can give you is that where Alias is very competant SW turns in reasonable karaoke with the occasional bung note to ground any illusions.

          I guess thats the difference between high end vs mid range. SW does a lot generally for a reasonable price and it is made simple to use.

          BTW I remember I once used the Alias manual to highlight just how poor the SW Help notes were from a technical info point of view.

          If SWv6 is based on the Catia core then surfacing should be better in the future.

          The practical questions are: how long would it be before it is as capable or better than existing SW, and how nobbled will the SWv6 feature be in comparison to how it is in Catia.

          I suppose you have some weighing up to do as to whether you stay with the DS stable in the lull of the transition or move on and up to something of full industrial strength.

          Perhaps for your type of work the shortcomings of midrange dont have any noticeable apparitions in which case there is no need to fret about whats deficient by high end standards.

          When you put it under the microscope even the cutest wee mite can assume horror movie proportions and look grotesque.

            • Re: Are SolidWorks surfaces good enough?
              Kevin Quigley

              I don't have any issues with SolidWorks for the majority of the work - and as the Alias guy said, you still need to do the production /engineering in something else, and for this Solidworks is the match of anything else. But it is for those projects where the ultimate quality of the surfaces do matter - the bigger scale items where customers are now going direct to tool from the CAD data we supply. The small scale items have done this for years but the bigger scale ones tend to go through a more staged development phase of full sized prototypes.

               

              There are  two issues at play here. Firstly more and more toolmakers have 5 axis capability now and the software driving it is better so getting quality A class surfaces directly is common place. Secondly, as companies gear up and get some in house capabilities these issues become more critical in the supply chain. It only takes a few calls and emails before we start getting our customers calling asking "why are there problems with your files"? This has happened in the past and is happening more frequently. 99% of the time it is the operator's lack of experience at the customer and/or suppliers end, but there have been occasions where we have handed over data that is OK on everything we have here, yet is creating issues at tooling (usually, I have to say due to ancient CAM software). In these cases the mite gets blown up to supersize then emailed to all and sundry and I have to spend time placating customers and their suppliers, and doing some education.

               

              I know these issues happen with all systems, and when we use a Solidworks only supply chain (us, the customer, the toolmaker) there are no issues, but this is rare. I would prefer to stay with SolidWorks for the whole process. Maybe I just need to get better at it.

                • Re: Are SolidWorks surfaces good enough?
                  Neil Larsen

                  Possibly, although an expert knowledge of SW surfacing is hard to come by simply cos they dont publish that very important inner workings stuff anywhere.

                  You would have to troll through years of forum posts and tuts to find some random gems of info.

                  Maybe you could post an example of a disappointing surface here for Mark B to comment on?

                  Of all the SW people he has been the most helpful to users in understanding the best application of the tools although I think he does so as an extra and out of pity for the knowledge and index starved user. I did once or twice suggest he should produce a series of master classes on the intricacies of surfacing but I guess the higher ups still believe fervently in the pervasive powers of intuition...

                  Perhaps the issues that have been 'surfacing' are a general limitation of the tools as they are coded or it could be that the way you went about it was more prone to lumps and bumps and waves.

                  I seem to remember a thread, probably years ago now, where improving the surface visualisation/evaluation and particularly along and across edges was pondered by Matt and yours truly. You would need to do a search..

                  That was when the SW mojo was at its peak though, and people felt they were participating in something worthwhile and being effective in bettering the program itself.

                    • Re: Are SolidWorks surfaces good enough?
                      Kevin Quigley

                      The problem or rather potential issue I see is not the lumps and bumps in the surface but rather the density of the points that make the surface up. Back in the day when I was a Think3 user (which I moved to as SolidWorks surfacing at that time was a bit limiting) we spent a lot of time on the WebEx training sessions looking at surface point density. Ask any boat designer or auto designer and the fewer the surface internal points the better. In fact this was one of the main reasons why Think3's Global Shape Modelling kind of died the death - every time you did a GSM edit the surface density went sky high. A few edits and the file size went up 100x and became unusable.

                       

                      That is the issue I am seeing. No matter how few the curve control points that drive the surface, the resulting boundary and (especially) fill surfaces have many control points internally. You cannot see these inside Solidworks - so you have no control over it (that I know of)- it is only by exporting to other apps that you see this. This makes subsequent editing of these surfaces impossible in other apps - you have to rebuild.

                       

                      I'll try and post up screenshots of the Alias surface and the SolidWorks surface as seen in Rhino tomorrow.

                • Re: Are SolidWorks surfaces good enough?
                  Kevin De Smet

                  You have to keep in mind a fundamental difference between Solidworks and Alias.

                   

                  Solidworks has a boundary representation kernel, Alias uses Nurbs math exlusively. That make Alias a heck of a lot dumber than Solidworks in handling complex surfaces - wholesail. That's filleting, shelling, all the usual but also on how to smoothly calculate a 100 span fill surface.

                   

                  I'm not saying it's good, or bad. Just that it is not apples-to-apples.

                    • Re: Are SolidWorks surfaces good enough?
                      Kevin Quigley

                      It may well not be Apples to Apples but what I am seeing is this:

                       

                      Solidworks.png

                       

                      That is the imported IGES file from SolidWorks (imported into SharkFX) - a boundary surface and a fill surface - with iso lines On and internal surface points on (neither of which you can do in SolidWorks).

                       

                      The image below is from Alias Design, as done by a Alias expert showing the area in the fill surface above.

                       

                      Alias.png

                      That little surface at the bottom left is a patch. The main surfaces are two surfaces. Note the dramatically fewer internal control points.

                       

                      My question is, if my curves are two point splines or minimal splines why does SolidWorks generate surfaces with hundreds of internal control points? If I was sent this data I would be looking at it and thinking...hmm...not so good.

                        • Re: Are SolidWorks surfaces good enough?
                          Rick McWilliams

                          I have some serious problems with airplane fuselage shapes. I use mostly boundary surfaces. There are often some nasty looking concave surfaces for shapes with entirely convex sections in both directions. Solidworks surfaces are often very much worse than they appear on the screen. We definitely need some viewing tools to determine the quality of surfaces.

                           

                          Solidworks files are full of crap. There is a bunch of obsolete data for features that were long ago deleted. There is a bunch of dead space and very verbose data structures. If the file were reduced to its essence the file size would be 1200 times smaller.

                            • Re: Are SolidWorks surfaces good enough?
                              Kevin Quigley

                              Some interesting points here. I use Ashlar-vellum Cobalt and Shark as well and I have to say there is nothing I can model in either of them that I cannot do in SolidWorks. In fact SolidWorks surfacing is more robust and allows you to build from edges of existing surfaces or surface groups maintaining G2 continuity (which neither Cobalt or Shark can do in most situations). But coming back to the point of my post, SolidWorks surfaces appear to have a very heavy internal point count whereas other apps internal points are governed by the curves that drive the surfaces. If you look at that Alias surface I posted the internal points match the curve CVs that drive it.

                               

                              I can kind of understand why the Fill Surface gives rise to a heavy point density - as this is essentially a semi automated patch command - but the Boundary surface should not do the same.

                               

                              Maybe someone from SolidWorks can chip in here and explain?

                                • Re: Are SolidWorks surfaces good enough?
                                  Kevin De Smet

                                  Boundary and loft surfaces are procedural, it's not "pure" Nurbs, it could be that the import into other applications is fitting it with very tight tolerance nurbs. Just a hunch on both the pureness there and of course blaming it on the import of other software, 'cause that's always easy to blame it on, right!

                                    • Re: Are SolidWorks surfaces good enough?
                                      Kevin Quigley

                                      That might well be the case, and if it is, it is even more of a concern if you are supplying data to a toolmaker who does not use SolidWorks. The last thing I'd want to see is my carefully sculpted low CV surfaces turn to a heavy uneditable mush in some other system.

                                       

                                      Then again maybe it really doesn't matter these days? I'm going back 10+ years when exporting files to a CAM system was always hit and miss, so you made sure the surfaces were as lean as possible.

                            • Re: Are SolidWorks surfaces good enough?
                              Roland Schwarz

                              There is a basic "language problem" between SW and Alias (and Vellum) modelling.  It is likely true for othe ID surface modelling applications.

                               

                              If you look at the basic structure of Alias and Vellum surfaces, they tend to be based on elliptical U-V curves.  For example, you can perfectly replecate many Alias and Vellum surfaces using variable section sweeps with elliptical sections.

                               

                              Another issue is that Alias and Vellum knit their surfaces with much looser tolerances.  I have seen this many times, where the calculated intersection of two surfaces in Alias or Vellum does not match a more accurately calculated intersection in SW (or Pro/E or UG/NX).  Alias and Vellum also allow a designer to "force" a highlight or intersection to appear far from where the theoretical intersection of surfaces would actually be.  UG/NX allows users access to the parameters that control the tolerance of parasolid edges.  There one can really see the effect of this.

                               

                              Alias and Vellum can model B-surfaces with more degrees than SW.  A typical Alias surface is 3rd degree in one U-V direction and 5th degree in another.  SW does not cope with this well.  (UG/NX does, and allows access to this information.)  Because of this, it can sometimes be difficult (impossible?) to create the same surface from scratch in SW that one could make with ease in Alias or Vellum.

                               

                              SW appears to have practiced a deliberate blindness to these and other issues concerning surfacing.

                              • Re: Are SolidWorks surfaces good enough?
                                eric fairbanks

                                Great topic and one I have also thought about for quite some time.  I would love to have access to deeper levels of surfacing tools in SW or at least some clarification to the points brought up in this discussion.  Like Kevin, I also run into the same sort of issues when I take models into Alias for analysis.  I would love to see imporvemtns and more control of surfaces in SW v6.. **fingers crossed**