9 Replies Latest reply on Nov 27, 2008 6:49 PM by Kevin Quigley

    surface knit stability

      So, I have had many issues trying to knit surfaces that should knit perfectly.

      Example 1:

      Copy faces of a solid (created in SW 2009)
      hide copies
      delete originals
      attempt to knit, failure to knit faces, with any option set.

      (and when this happens, replace face crashes SW)

      Example 2:

      Knit feature yields a solid, but if I go back in the tree and change something elsewhere in the model, the knit fails.

      Example 3:

      Imported geometry, forget it.

      Example 4:

      Knit feature yields a solid, do a verified rebuild and knit fails to rebuild.

      I know what you are thinking....Perhaps I have invalid geometry? Well to me, if it worked once, twice, three times...It should work again! Ya....Do an entity check! Well, SW crashes when I use check entity...

      In examples 1, 2, and 4, all faces will knit to adjacent faces, but will not knit surface body to surface body. The only solution I have found is to manually select each face, which can literally take hours, and it might not even work.

      I spend about 2 hours per day dealing with this issue.
      I am seriously pissed. My VAR is so far incapable of helping me solve this issue.
      Yet another reason to switch to think3!
        • surface knit stability

          Perhaps you can post some files to the forum so that we might be able to shed some light on these issues.

          • surface knit stability
            Matt Lombard

            Yeah, you might see stuff like that and a million other problems of the same kind. So what? It's the cost of using SW for surfacing.

            Anyway, if you are modeling with verification on rebuild turned off, and then you're having knit problems, you need to just keep it on.

            There are some things you just shouldn't expect SW to do. You need to develop an intuition for what type of stuff SW can do and what it can't. Some models do this kind of stuff all day, and some don't. To be fair, there is usually something you can identify that is at least questionable when it doesn't work, but not always.

            As Mark says, you really should post some parts. Then we can get an idea if you are running into real issues, user issues or training issues. Sometimes a different approach fixes stuff you don't expect. Surfacing is not as predictable or reliable as solid modeling.

            By the way, I agree that people who do serious surfacing should be looking at other tools that have that sort of work more squarely as their focus.
              • surface knit stability
                Leonardo Sanchez
                Interresting that you mention Think 3.
                I used it for about a Year (allthough in 2001 ) and while it does
                have some impressive surfacing (especialy GSM ), it had me aching for
                the SW-scetcher and the general SW-workflow.
                I still wish I had it ,but I would'nt want to be stuck using it exclusively.
                High -end surfaces and efficient solids in one app may still be the terrain
                of Catia NX etc.
                  • surface knit stability
                    John Kreutzberger
                    I cannot speak directly to the initial poster's issues. However, I am going to say that a small tightening of the tolerances SW uses when knitting would not necessarily be a bad thing. In my molddesign work, I use surfaces for creating fancy shut off surfaces and parting lines. I just did a fairly complex tool in 2009 (based on imported geometry) and everything knit as expected. `as expected' being the key phrase.

                    In past versions I felt that SW was bit too forgiving when knitting and sometimes would distort adjacent surfaces to force a knit to happen. This week will mark the 10th anniversary of my switch from cadkey/fastsurf to SW. I have seen incredible improvement in surfacing capabilites, but naturally get frustrated as well. I will keep an eye out for the sort of problems described here and post accordingly.
                      • surface knit stability
                        Ahh Matt, ever the pragmatist!

                        Your advice is well directed. I am aware that I am pushing SW further than it wants to go, yet occasionally it seems that the issues that arise are simple enough to be solved effectively, but this is not so.

                        Still, I am only using the tools given to me by the software. I feel like there is a carrot on a pole just out of my reach. What is even more frustrating is when the carrot is momentarily within reach, and then taken away once more. Surface knit and replace face have been around for a few years, I'm still waiting for them to work as advertised. Sometimes they are a treat, other times I want to switch software.

                        I feel as if I have had a good intuition of what SW will do and what it won't, having been a regular user since '98, doing complex surfacing daily since '02, but this latest release is completey irregular. What used to work sometimes won't anymore, and things that were not possible previously can now be done easily. Problem is, in the 20 hours per week that I spend dealing with failed surface knits, I still can't find any reason why some knits fail. Many times I can find a solution, but usually the knit that should work doesn't and I can't figure out why. Most of these fall under example 1 in my OP.

                        Of course, my VAR is able to knit perfectly every time...

                        As far as posting files, I'm limited by draconian non-disclosure red tape, so jpegs are about the most I can post. (how draconian? a coworker just got reprimanded by senior management for having a competitor's sticker on his bike!:confused
                          • surface knit stability
                            Matt Lombard

                            Will Smith wrote:


                            Ahh Matt, ever the pragmatist!

                            Ok, hope that is a good thing.

                            Anyway, one thing I do to make sure surfaces knit is make surfaces always larger than they need to be, and then trim them back with a mutual trim so they are immediately knit. I've even seen this work on something as stupid as simply extending a surface, and then trimming it back. You might at least try it.

                            You have to take JK's advice seriously. I've seen several examples of the "black hole" surfaces, which suck everything down to a point, distorting surfaces around it. I have a feeling this has to do with SW forcing stuff that falls slightly out of tolerance. Making tiny changes usually fixes the situation, or causes it, depending on how you look at it.

                            The tool has its limitations, all you can do short of switching to another tool with other limitations is work around them. And maybe scream a little, but working around them is more productive.
                              • surface knit stability
                                John Kreutzberger
                                Hey-I was just knitting some surfaces and saw a new checkbox in the knit dialog in 2009: minimal adjustment. I wonder if this is what I'm looking for as far as not tweaking surfaces to get them to knit. I'll play around with it when I get a chance and report back.
                                  • surface knit stability

                                    Sorry to disappoint you that this enhancement is not what you think, but it is an important enhancement none the less. We basically put this in to overcome tangent edges automatically healing when it was not the designer's intent to do so. Basically, pre-2009, split curve features, which were intentionally created (part lines, style lines, comolding edges etc.) would be eliminated by the knit feature (child feature to the split curve edge.) Minimal adjustment ensures that these features are kept.

                                    Will, still waiting for you to post something. If you have propietary issues, post an example.
                        • surface knit stability
                          Kevin Quigley
                          Think very carefully before going to Think3. I used it for two years before they upped the price (doubled it in fact). As Leonardo says, the sketcher is rubbish, the history is flaky at best and getting EXACTLY what you want in surfacing is not actually that easy. The GSM stuff is impressive, as is capping but don't expect your history to stay intact. Not sure what the pricing is like in the USA but here it is well into the CATIA price bracket, and to be frank if I was inclined to spend that kind of cash well CATIA just gives more options IMHO. But the biggest flaw I had using it was that the features in solids rarely "took" on complex geometry without loosening off the tolerance.....and as we all know that creates merry hell when it comes to export time!

                          Like a lot of applications I think Think3 is best suited to editing geometry created in another package as a prep for tooling. If you want something similar but at less price take a look at VX. Thats what I moved to after Think3. Not much you cannot do surfacing wise in that package (and it also has some elements of GSM using the VX Morph tool. It has the same open solids/surfaces as Think3, and the IGES translator is top notch.

                          But....fact is, I still use SolidWorks for 95% of the work because it is just so much more reliable. History is rock solid, and more importantly, very east to correct if things do go astray. If SW had conics, and easier spline manipulation in 3D I'd probably use it for everything.