More Detail on the GW3D Blend
I wanted to add some detail about one specific feature, the Blend. I started with this Boundary surface in Solid
Works, which is intentionally a little wobbly. This was created with 4 splines, all with a sharp bend in the middle. Then I trimmed out the corner with the intention of replacing it with a Blend.
The imag
e on the left is the PropertyManager for the Blend Surface. It’s long, and a lot of the panels expand, depending on options selected. To be honest, I’m not sure what all the settings do. It’s not a great thing to evaluate something if you don’t understand it completely, I admit, but I’ll be upfront about that, and just give my observations.
I don’t completely understand the “spine”, but it appears to be like the centerline in a centerline loft, where the spine doesn’t really touch the finished geometry, but it guides or influences it.
The Direction/Flow options are Normal To Curve, Least Tension, and Relative To Chord.
The Continuity options are G1, G2 and G3.
Scale factor I presume has something to do with the percent distance over which the curvature of the blend is influenced by the other surface. So a G3 connection with .5 Scale turns out to look sort of chamfer-ish.
The other consideration is that higher orders are not necessarily better. If you think about the progression from line to parabola to x^3 curve to x^4 curve, the higher the order, the lumpier the curve. This is why conics have the great reputation they have – they are essentially the parabola, which is predictably smooth, and never flips convexity (doesn’t get lumpy).
If you compare a SW C2 Boundary surface with a GW G3 Blend, the Blend can only be characterized as lumpy. Now maybe with more controlled surfaces like stuff created from controlled conics this situation does not stand out as much, but here it does.
Using this as the only test is a little unfair, but these are the situations in which I would find myself with this software. A completely thorough test is unrealistic. After XX years of using SolidWorks, I learn new stuff about what it will and won’t do every day. You might be able to set up special conditions under which each of these tools perform ideally. I rarely run into ideal conditions.
I was not able to get the GW G3 surface to knit with the other two surfaces. The GW G2 surface does knit.
I’m not sure why the base surfaces change a little between the curvature images. They should be all the same except the blended strip. The accuracy of the Curvature display in SolidWorks is pretty questionable.
For the G3 condition to be better behaved, I think it needs more control. Just giving it two edges and some boundary conditions doesn’t do the tool justice. GeometryWorks certainly allows for more control, but in this simple situation, it doesn’t perform as well as the simpler solution.
I added a draft to one surface to mix things up. The blend surface looks nice. There are too many options. Zebra stripes look good. The basic geometry will force something like a flat spot. A very smooth flat spot.
To make it more simple, I tried the blend curve in GW. There are six different types: cubic, rational, rho, least tension, curvature radius method, and curvature linear method. In addition there is selecteable G2 or G3. The curvature combs all look reasonable. The first 4 types are only tangent matching. G2 and G3 curvature combs seem to do the right things.
I have only used blend surface a few times. I do not really know how to use it. I will add some images tomorrow.
I just tried this in a dedicated surface modeler and had the same results, I think I’m doing something wrong here. I would expect to see a nice continuation in the curvature plot but it seems impossible with this setup?
The end points of the two splines, and the tangent vectors at their ends seem to be pointing towards eachother almost straight. I suppose that’s the problem. but then how do I get the G3 combplot that I want???
This is two accelerating splines extruded in standard Solidworks, notice I tried tweaking the tangent length on one of the boundaries of the Boundary surface but that made it not even C2 anymore. Hmm.
Wonder how GeometryWorks handles this?
[img]http://www.dezignstuff.com/blog/wp-content/uploads/2013/05/swsrf1.jpg[/img]
[img]http://www.dezignstuff.com/blog/wp-content/uploads/2013/05/swsrf2.jpg[/img]
[img]http://www.dezignstuff.com/blog/wp-content/uploads/2013/05/swsrf3.jpg[/img]
Hmm ok then Matt can you please try cutting away a bit more of the surfaces and see if it cures it. I am interested to know 😉
As far as I am aware, C2/G2 means the radius of curvature where the 2 curves meet are the same. C3/G3 means the rate of change/acceleration of the curves is also the same where they meet. This normally means you need more “space” for the curve to flow smoothly and probably why in this case the G3 surface is lumpy – the source surfaces have more influence over the new G3 surface and the software is trying to shoehorn in a smooth surface in too little space.
Hello all, i really REALLY appreciate a REAL example of how to use this tool with C2 or C3 application, as neil and Rick explain on the fusalage of an airplain, but when to use it? what would happen if instead of using C3 is only possible to use C2 or vice versa? thanks so much
@Rick McWilliams
“The twist axis is located near the hinge line.” Interesting, great explanation.
Yeah OK so does it have those undulations for all G3 cases? If so its more or less useless. Are you sure you are using the right settings? If G2 resembles SW boundary well it really has no advantage. So basically then this is only well behaved for things like Ricks wings,fuselage. Is that it? Disappointing… kind of kills my interest off… Probably these standing waves are present in the multiblend case as well. I guess I need to download a trial and judge whether it does enough well enough to justify the price. I do have something like Rick is doing coming up and where the fairness is going to be quite important but….
A tapered wing tends to load the tips. The twist is called washout, it used to adjust the loading so that the inboard part of the wing will stall first. The twist axis is located near the hinge line. The aileron hinges can then be on the surface of the wing.
@Rick McWilliams
Twisted…is the Trailing Edge twisted up at the end of the wing for low speed stability?
I use the GW blend surface to make a ruled surface for a tapered, twisted and cambered wing. It does not produce mid span sags.
[img]http://www.dezignstuff.com/blog/wp-content/uploads/2013/05/wingblend2.jpg[/img]
Ok, here’s one that is set up to work. Nice smooth surfaces, and I’m using the Blend surface. At default settings, this severely egg cartoned. I dialed back the settings so here you can see the beginnings of the undulations.
I don’t mean to rip on the software, but this particular feature isn’t one of my favorites.
Kevin, the pre-curving is one way of dealing with c3. Another is just to plan a larger space for a c3 blend vs a normal radial fillet.
Matt, could you cut a curvature comb plot across those surfaces? It’s very important for G3 transitions that the base surfaces start to accelerate a little before the blend starts, this makes the G3 transition so “legendarily” smooth.
If you don’t do this, and just tick on G3 that’s an option in a box (ok ok, dropdown list… to be precise, lol) then you’ll get the wobbles and horribleness you got there
@Devon Sowell
Yes it is. Tools like Power Surfacing and T-Splines are attempting to change that in CAD, but even those are not really an end-all solution since they are not good in a lot of situations. And even when you do have use for sculpting a shape, it still takes quite a lot of time and skill to get the form exactly how you want it. It is not as quick and easy as the slick demo videos make it seem. It takes a lot of practice.
GeoemtryWorks is difficult too. The main benefit of it is simply that it gives you some alternatives to the base SolidWorks tools.
High end tools like NX and Catia have some more controllable sculpting tools too that can make surfacing a bit easier. But they are such huge and complicated programs that it takes a lot of learning to understand what to do. Relating back to another post on this site… I guess that’s why we all have jobs, because it’s not easy, or cheap.
I’m an intermediate CAD surfacer & always will be.
From my view CAD surfacing is very difficult, very complicated, & very expensive.
If you scroll down there are some short presentations/pdf on this Japanese webpage http://www.t-sol.co.jp/GW3D/GW3D_top.html in Japanese though. Perhaps these are available in English somewhere but I couldnt find them or much on Youtube. Unfortunately my Android tablet doesn’t like the GeometryWorks website because of Java 🙁 Seems it sells in Japan for ¥198000 which converts to US$2000 currently. Not sure how accurate that is… perhaps that’s after money printing.
BTW does this add in work properly with the Freeze Bar?
I think you are being a little unfair by making your surfaces deliberately messier/bent/wobbly than a competant user would use as a basis of their work. Sure there might be rather a deliberate wave in the splines but I think you should assume those would be smooth in themselves. Show me how an Ed Eaton ‘atomic bomb’ fillet stacks up with conics, say. Perhaps Rick would post us some simple examples of situations where GW3D excels seeing how he is more familiar with using the program. Maybe he can use his plane to point out the basis of the construction and which feature he used.
The accuracy of SW curvature display is something I have asked Mark B. about a couple of times before and never got an answer. Probably he will say he cant remember it or he didn’t understand the question. I guess it must be embarrassing to talk about.. Maybe its one of those company secrets like the price of products.
CAD really still does have its foot planted in a bygone business era in many regards. If they shift weight to the other foot it will be because they are confident they have it planted in something else overly protective like rent seeking via the cloud. I wonder if they realise how ridiculous this is viewed with objectivity. Possibly ex CAD vendor employees are best suited to selling vacuum cleaners door to door….the kind that are overpriced, have to be sent back to the factory for service and you can only get the bags for from them on subscription. How many suckers are there in this story?
I have only used GW blend surface with nice end conditions. It will do a ruled surface with two planar curves. It will do a beautiful smooth fillet between surfaces.
The multiblend is even more complicated and it requires suitable end conditions. I use it on the aft part of the engine cowl. Conic surfaces also make very nice fillets.