Need Help with the Boundary Surface Tool

Hello Ben,

Not sure who, how, what, where or why but there are some definate mis statements there, if what was said is true. You should most certainly read through the help topics on the Boundary surface to understand how the feature works. What the underlying functionality is doing behind the scenes. But just to help clarify, from the first initial look in some ways they look like they perform or create the same surface. Some differences that you will notice:

1) Controls that you have at the end conditions (Are at least equal if not surpass the way that loft tool works) 2) Not as much sketch geometry is needed to define the boundary. 3) There is a Hair Drying video on the Solidworks main web site that shows some of how the function works.

You should also check the Dimonte group web site. No one lays the smack down as well as Ed with his tutorials.

I dont agree with your VAR. The boundary does not always give you the smoothest surfaces, especially when you start changing tangency values. I believe that this is why Solidworks included it as a separate command, though they would have liked to have just improved the Loft function.

Boundary allows you to create a surface with only a certain subset of a long edge, and a curve that does not have to lie at either end of the curve ie you can grab the handles and slide them along an edge. It extrapolates design intent pretty well. Boundary also allows curvature continuity along what would be considered guide curves in loft. I use Boundary extensively, but substitute lofts when the boundary solution looks crinkly.

I agree with Thomas. Plus, a couple more comments:

Mark B has been trying to get people excited about the boundary surface. It really does have potential, but it isn't a regular in my tool set yet. It fails too frequently, mainly when it thinks that the tangencies from two different directions cannot be reconciled at a corner. Very frustrating.

I find the interface of the boundary feature difficult to navigate. It takes way longer to set up than the Fill surface, and those curvature combs! You can't see anything when they're on, and they're on by default, and don't remember your last setting. All the colors and combs and stripes and whatnot on the screen make my head hurt. I could live with a cacophanous interface if the feature delivered in the end, and it does deliver about 20% of the time. I prefer to use Fill which is far more reliable, works in a wider range of situations and is far easier to use. I have had isolated instances where Boundary gave a better result than loft or fill, but that was after a lot of trying. You also have to tweek the options to get the best result.

The one word for the Boundary feature I have is: inscrutable. I don't know when it will work, or why, or what settings it will take. I'm optimistic and keep trying, but I have more confidence in Fill. Mark's a good guy, but he is over confident in this limited and frustrating tool.

The same thing happens with Loft with Guides. With Loft we do some "fudging" but when it comes down to it is not any better (turn on display>curvature) than the quality of Boundary. In addition, many users that are successful with Loft, don't use guide curves because they've figured out that, like Johnny Depp says in Pirates " Well they be more like guidelines than rules". Users know that you can not count on the accuracy of Loft edges that were created with guides because they don't follow them very well. Think of them more as influencers for the direction between the profiles. This is one of Ed's main points when he teaches about Loft, and he's correct. The reason that Thomas and others (me included) are successful with Loft sometimes is for this very reason: because of the influence profiles have over guides. In some cases this works better. This is the key difference between Loft and Boundary: Boundary's 1st and 2nd direction curves have equal influence.

This is not true. The settings ARE remembered in the session of SolidWorks. If you feel they need to be remembered as a doc or system setting, then please submit an enhancement and I'll follow it up. It's not difficult to change.

All the colors and combs

I can show you dozen of examples where Boundary outperforms Loft in quality and speed. The fact of the matter is that we just didn't just dream up a new feature, but we need to because of the limitations of Loft with Guide curve. In fact we did an extensive benchmark in 2005,

2006 with hundreds of customer parts - solving there loft problems with the Boundary feature.

I don't know how you can even compare Boundary and Loft to Fill. Don't get me wrong, Fill is SolidWorks' "Crown jewel", but it can not substitute or surpass what Loft or Boundary can do because of its limited internal curve influence.

Matt, I'll have you talk to my wife - she'll tell you about the real Mark - flaws and all

Seriously - yes I am confident in Boundary for what we created it for

- to overcome the bias influence of profiles verse Guides because users wanted a "true" boundary feature (like ProE's boundary Blend or ISDX) and they were trying to use Loft with Guides to do this. Boundary uses the Gordon Surface algorithm which is more accurate and more efficient (especially when using many curves in the network)

What we might be seeing when comparing Loft with Boundary, is that Loft is more "forgiving" but the "dirty laundry" is still there - especially when performing downstream features. The Boundary feature is more exacting, and demands from the user, better sketch curves and edges in order to attain a better quality surface. Perhaps this is an issue we can look into further to see what we can do to make Boundary more forgiving but we do run the risk of assuming what the user wants as a result; a result that might not be the quality that they need.

Great topic and great discussion - I realize that there is still much confusion about the differences between Loft, Boundary and Fill and was the reason I dedicated a breakout session to it at our last SWW.

Regards

Mark

Thanks, guys. I'm a little less confused than I was before this discussion. Sure wish I had been able to go to SWW!

Jerry Steiger Tripod Data Systems "take the garbage out, dear"

Mark,

Any chance that sort of thing can make it's way into the documentation? Should I be looking for it on the Customer Portal instead?

I have only played with this on someone elses pc since I'm on 05 but I would much rather you applied quality to boundary than made it compliant. The whole reason I wanted continuity improvements to edges was that I didnt want to obsess about cutting and patching like Ed does - sorry Ed - I see that as admireable but very unproductive. I think they have complimentary roles.A tool to achieve your intent with out being too fussed about conditions and another to join em up well. As for the curvature combs...well a lost cause sadly

Mark,

What I am seeing is that in a corner, if the Boundary cannot match the tangencies between the two different directions, it just gives up, and gives you nothing except a friendly error message. The feature cannot be created at all. What is most frustrating about this is that you could trim a rectangular patch out of a single surface, and still run into the error of matching tangencies between the two directions, which doesn't make any sense geometrically. Rarely can I get the curvature option to work in either direction. Sometimes there is an advantage to getting something that is somewhat fudged in the corners rather than nothing at all.

Here's an example.

This is a simple revolved surface, where a rectangular cut out is made in the surface, which is then filled with the Boundary surface. If you were setting up an easy slam dunk for this feature, what else could you do? Turn on Zebra stripes and notice one side of the existing patch is not even tangent. So edit the feature and set the first edge in the Dir

1 edges to Curvature. Notice the error that you get, "The guide or profile curve's curvature at its end point(s) does not match the curvature of the surface it touches." All that is selected here are simple edges created from the trim. Interestingly, it still works, but usually when I see this, the feature will not even work.

Look at the 3 configs in this second part. Loft can't even be created with c2 on the profiles. That stinks. Boundary works, but with the error shown above. The best looking patch is certainly the Fill.

Boundary gets compared to Fill because fill works in any situation where Boundary works (except, as you mention below when internal control curves are needed), and many situations where boundary doesn't work (such as non-4-sided patches). Fill seems to "fake it" sometimes too. Early in the 2007 release, you kept giving warning symbols on fills with sharp corners. That doesn't happen any more. I assume that the approximation it was giving the warning about hasn't changed, the symbol has simply been taken away.

No, Mark, it is true. It doesn't remember the setting from one session to the next. Because I don't generally make multiple Boundary features in a row, I tend to do it at widely separated intervals, between which SW has been turned off and restarted. So every time I use the Boundary, I have to turn off the setting. Curvature combs with no caps. Freeform doesn't do it that way does it? Loft doesn't do it that way either. Why does Boundary automatically turn on combs? Let me turn them on if I want to, but the default should be off.

Really, I'd like to see a few of those. Not that I don't believe that it works, I know it does from the once or twice I've gotten really good results from it. I just want to see what kinds of situations it works well in, and what settings you used to get there. I at least try it whenever I have a 4 sided gap to patch. I keep trying mainly on the strength of your optimism, certainly not based on my own experience.

Yes, you're right. If I'm using Fill, it is in places where I don't need to control internal curves. On the other hand sometimes you just need to blend over a patch without any internal control curves. In these situations, you can use either of the three types. I have often used them all and compared the results. Loft is the most predictable, but lacks GC conditions, as you mention. Fill is the easiest to use and works in the most general cases. Boundary remains tempermental and sometimes simply refuses to produce any results at all. I think I sent in a tech support problem wrt boundary not working a couple of weeks ago.

I don't wanna know. I'm happy in my illusions. In my head, I imagine you as a nice guy. Nothing from knowing you has contradicted that. I'm not complaining about you, just a little disappointed with this feature. I see what it is trying to do, and recognize that it could be useful, but in practice, I just haven't seen many situations where it works. Provide some of those examples for us all. You're trying to sell this feature. I'll buy if I can see it work in situations I run into.

Right, I agree.

That is clear. So if it doesn't get what it wants, it just shuts down. Also, I have to tell you that the only times I have crashed in the past week have been when making adjustments in the Boundary feature. Of course it's nothing I can spoon feed to tech support, but there is a problem in there somewhere.

I don't know if "forgiving" is the right word. You can make the surfaces in SW, and make them with a loft using c2, and then the quality of the c2 condition across the boundary between faces is not good enough for the Boundary surface, and it just fails. It is complaining about the quality of the surfaces made by SW. Next time SW is footing the bill, I'll take the time to track this down, because I see it regularly.

Also, I've never had good success with the handles in the Boundary surface. Handles in loft are touchy enough, but in Boundary, they simply don't work whenever I've tried them.

I agree. I want to see examples of where it has been applied successfully. I think in general we agree about everything, except you have several examples of where it works, and I have one or two, and those were found only by trying all of the edge condition options and combinations with the curve influence options. I've got to say that those are both a major interface pain. There is no feature in SW that requires so much clicking around.

If you would like to say more about the curve influence options, and what they are really meant to do, I'd like to hear about that. To me, it is just a toggle for two different solutions - I don't understand the difference between the settings.

While we're at it, why don't you add a "merge result" option, like the Fill surface?

It is possible to make them tangent by selecting "Tangent Influence" under "Dir2 curves influence"

That error message is confusing.

John Layne

well there you go matt, rendered butt naked again :o)

Not sure about that

Deviation Analysis comparison for Surface Fill and Boundary.

hmm well that shows boundary to be much tighter than fill doesn't it? if I read that correctly boundary has less than

Tangent Influence and To Next Curve seem to have the same effect. Do you have any way of knowing before you try it how these settings are going to affect the geometry (other than Random Button Pushing)? Using the Curvature setting for the Dir2 edges should be what makes the connection smooth. The curve influence settings in the direction1 have minimal effect, and you can never tell until you finish the feature and run a Deviation Analysis. This is part of the hassle of using this feature. Deviation Analysis is not parametric, and has to be run after a knit, so changing settings, rolling down, doing the analysis, it's kind of a tedious process.

Thanks for showing that, but my confidence in the feature is little improved if the only way to find the solution is to pick through all the options one by one until you come up with something that works. I guess I'm looking for something deterministic, predictable, and a little less random.

John,

Thanks for the response. I guess I kind of agree with FlowerPot, although I'd say it differently. How did you come by that solution? Were you just trying random things? I would think the Curvature setting would determine the results of the deviation analysis.

Matt

Wow Matt, I'm going to have to respond to your response to me over the weekend, but great feedback and to the others that are in on this thread. I want to look at your parts but I suspect that its the surface trim that is causing the problem. Have you tried it in '08? We are aware of a nasty result of surface trims, and deriving a projected curve from it, and have been working to resolve it in '07 SPs and '08.

Yes, display analysis in Boundary does not transfer from session to session but only in the session (as I stated). I would submit an enhancement (and other here that feel passionately about it (and qaulify it from a customer's perspective in the enhancement) to make it a doc or sys option - Power in customer hits!

I'll get back to you this weekend Matt.

Perhaps I should put together an interactive webcast on the subject of Boundary vs. Loft vs. Fill or do it with a select group of you. In any case there is a lot that we could learn from this discussion.

Regards

Mark

Excellent discussion guys! Lots of good information. I haven't had as much seat time lately with Boundary as I did towards the SW2007 beta and release. Overall my feelings on it are that I try to use it first, then revert back to Loft if I cannot get the results I'm looking for. I have had some pretty good success with Boundary in general but I would prefer to use it most of the time since I feel this will be the tool of the future. I need to crack the toolbox open on it again in SW2008 and test some more. (I need a few extra hours in my day it seems :-) )

Mark - I think a webcast on this subject would be great. I would certainly be there for it if my schedule allows!

Best Regards,

Ricky Jordan

I am not sure this is a good discussion.Discussions here are rarely of any real worth and interactive webcasts even further down the list let alone exclusive ones. I think it shows that people really dont have a good grasp of what the tools are intended for and to this extent SW have failed to communicate the differences or perhaps how the rules/purpose have changed. This is not so surprising however as the help notes have always been a list of features rather than educational about function or their mathematical basis. Matts example is in my mind is not a good one anyway. If you have a large sheet with a hole in it you are not going to use fill or boundary to patch it. The matter of what happens between sheets is more important. Pointing to very minor deviations in this example and saying the feature doesnt work is a bit inappropriate. How

Mostly Random button pushing :)

Having Zebra stripes turned on certainly helps narrow the odds very quickly (using the toolbar Zebra Stripes not the annoying boundary only version in the command, I'd rather have Zebra Stripes on for the whole model, shame you can't turn this on and off within the boundary command, you have to set it on before you start which can make it hard to pick sketches) .

For me, at least, Surfacing as whole can be "random button pushing" lofting in particular e.g. if something doesn't work try something else. After a lot of experimentation and experience I get a "feel" for how a loft will react but I can't always predict it, when guide curves are involved.

To quote Mark B: The same thing happens with Loft with Guides. With Loft we do some "fudging" but when it comes down to it is not any better (turn on display>curvature) than the quality of Boundary. In addition, many users that are successful with Loft, don't use guide curves because they've figured out that, like Johnny Depp says in Pirates " Well they be more like guidelines than rules". Users know that you can not count on the accuracy of Loft edges that were created with guides because they don't follow them very well.

I have found the Boundary Surface command very useful recently, it solved problems that I couldn't with lofts or surface fill, (that's not to say someone couldn't have solved the problem using lofts or fill). And in combination with the "Atomic Bomb for fillets method" it has proved very useful.

Look forward to that presentation Mark.

John Layne

Now THAT would be sweet. Would you mind observers? (always willing to learn and I'd be quiet in the back :-) )