Surfacing Top 10
Dan made a great point. If SolidWorks isn’t going to listen, let’s just talk to someone else, about things that interest us, no less.
So, for all you surface users out there, let’s do another Top 10 list for what kind of stuff you need to do your work. You can do one for what you would need SolidWorks to add, change or fix, or just make a list for what a new package would have to have to be your dream surfacing software.
If I were looking for a new CAD software to do surfacing, these are the things I’d be looking for:
- ability to create, evaluate and control curves effectively (splines in 2D and 3D, C1, C2, C3), this would include helix, eq driven curves, conic curves, b-splines
- full hybrid capability – meaning that solids and surfaces should as far as possible be treated the same, and this implies extensive multibody capabilities.
- direct edit capability (push/pull CVs and other controls) on general NURBS surfaces. Like the fancy stuff in NX, Catia, Solid Thinking, Tsplines. The better this capability is, the less reliance there will be on individual feature-based geometry tools.
- A boundary feature equivalent (curves in arrangements like X, T, E, F, H, 4 sided, 3 sided degenerate, 2 sided degenerate), with C2 on all sides, and weighting. It would be great to specify options like minimum curvature, no inflections, and to be able to control the UV flow with equivalent of SW connectors (only smarter…)
- A Fill feature equivalent – basically N sided patch with unlimited N, and C2 on all sides, control curves, specify no inflections, etc.
- Fillet/blend capability. Suitcase corners, c2, variable radius, algorithms to find largest working fillet for a selection
- All the tools, knit (with gap tolerance), trim, trim by UV, extend, untrim, make solid, unknit, delete face with healing
- A genuine Ruled surface that works from edges or sketches, and you can specify the ruled line in various ways
- Shell and approximate shell – if you can’t technically offset a face to make a thin wall, then fake it somehow
- Capability to work with mesh data for reverse engineering from scanned data, STL files, or OBJ type meshes. This would include Tsplines equivalent functions as well as the ability to lay sketches over the mesh.
If I were to want SolidWorks to make improvements, these are my top 10:
- All surface and curve creation tools allow me to specify a maximum curvature value.
- Either make conics available, or allow me to specify “no inflections” on splines.
- Splines would never need to be “relaxed”. When splines get kinked up, you can have 2 different states of the spline without any of the control points or handles moving. This is dangerous, and has cost me money.
- When something won’t work, specify visually where it won’t work. “Fake it” capability for Shell and Fillet – like a wet thumb covered in Bondo.
- Make the Freeform feature into something other than a bad joke.
- The Fill feature tends to oscillate badly when you apply C2, also the interface for setting edge conditions is awful.
- All surface creation features should have the ability to knit the new body into the neighboring bodies.
- Direct edit capability (push/pull CVs and other controls) on general NURBS surfaces. Like the fancy stuff in NX, Catia, Solid Thinking, Tsplines. The better this capability is, the less reliance there will be on individual feature-based geometry tools.
- Additional Ruled Surface options that don’t get all wonky. It would be great to have options to simplify instead of complicate – for example keeping the straight line in the ruled surface parallel to a given direction rather than pivoting around that direction.
- Capability to work with mesh data for reverse engineering from scanned data, STL files, or OBJ type meshes. This would include Tsplines equivalent functions as well as the ability to lay sketches over the mesh.
Great topic. For tweaks to SW:
1. Face IDs must be absolute, non-duplicate, non-confused (ever)
2. No-flip surface directions/vectors (ever), so directions of surfaces must be known and controlled
3. As a result, I’d expect to see no-flip surface features (like Trims)
4. 3D sketch fillets will work, and be editable, without trashing my other curves
Mentioned stuff I really like:
Capability to work with mesh data for reverse engineering from scanned data, STL files, or OBJ type meshes. (Matt)
Geometric features can be organized by me and grouped into folders with sub folders. Features do not own or consume curves or sketches. Parents are always ahead of chidren. (Rick)
Something I’ve seen more and more often in newer versions of SW is things changing (unpredictably so) with no edits. The “relax spline” thing between versions is a particularly costly one, since you should have the ability to trust your software not to change/trash your geometry when you’re not looking (even between versions). But last week I saw something truly horrendous, and just happened to catch it when making other edits. I had a regular cut of a solid “to surface” that was an arbitrary distance from the surface. CTRL-Q didn’t resolve the issue. I had to literally edit the feature and then the preview snapped back to the plainly-selected surface, rebuilding correctly. How the hell does that happen? Really. Ramifications could literally be as high as the tooling cost (or worse, another order of magnitude if incorrectly-molded products hit the market a la “Spasmodica” from Elizabethtown with costly post-release recalls). No wonder they don’t want me designing planes with this stuff.
Yeah well I suppose resellers have a vested interest in ‘flogging’ SW for as long as possible…or until the DS timetable dictates. VARS have a dependency rather than an enterprise.
Its a bit of a pretense/deception though to be gathering ideas for improving SW and also assuming to relate to customers like the old magic still existed.
Actually I think the spell has been broken not only because customers wised up but because DS have pulled up the drawbridge.
Personally I find it a bit disturbing that new people are probably buying into SW unaware of the circumstance but…
@Neil
He probably just wants to get some other users behind some of the ideas. I agree it’s kind of futile, which is why I didn’t do it myself, but he’s a reseller, so he still has some optimism, and you can’t just flog it out of him.
@Kevin Quigley
Kevin,
I find that your comments cut close to the core of the SolidWorks / Dassault issue of the allocation of surfacing tools. Of course all SolidWorks users would like to have state of the art surfacing tools. But I give you credit for recognizing this as a differentiating aspect of high and mid end CAD systems. And you quantified this as a $5,000 base price SolidWorks package vs. a $20,000 CATIA package. Most of us SolidWorks users would be delighted if we could have the surfacing capabilities of CATIA, but pay no more than our SW buy-in plus maintenance upgrades.
From a strategic business viewpoint, Dassault cannot just give away surfacing CATIA style. All of the accountants in the multinational firms that buy CATIA, would love to get that capability in SolidWorks for 25% of the price. And this would be at the expense of Dassault’s bottom line and would literally decapitate CATIA.
That said, I think that the solution that you are hoping for is close at hand. It may not be in the form of the SolidWorks model that you have already bought and paid for, but will be offered to you in other forms of poison:
1. Autodesk will offer a package of Inventor/AutoCAD/T-Splines/++ for about $15,000. I think that T-Splines gives them the ability to provide an offering that competes with CAT
IA and NX, which is much more profitable than just slugging it out with SolidWorks. With T-Splines they may have the ability to undercut CATIA and NX pricing by 30-40%. This gives them new market and growth that they did not have before. I find it hard to believe that they will use this to capture SolidWorks market share.
2. Autodesk might also offer a drop in version of T-Splines to SolidWorks user for about $4-5,000 which would definitely carve into Dassault’s CATIA base and SW too.
3. Dassault will have to offer an upgrade path to CATIA which credits your investment in SolidWorks. Maybe this would sell at a $10,000 upgrade????
4. SolidWorks will sell CATIA level surfacing “on-the-cloud” for about $20.00 per hour. Just work fast.
5. Wait a few years and surfacing technology will be yesterday’s technology and even SolidWorks will include it free. I think that Dassault’s biggest problem is moving CATIA’s rock up the hill. 3D CAD has so far to go to be truly powerful, there is a lot of development necessary, and someone has to be found to pay for it. Surfacing will be just a “used car” in the near future.
Steve Jobs may have been maniacal, and controlling, but Apple has a lot of happy customers. For sure SW2012 does not have a happy customer base. But, SolidWorks is much like Apple ’94, except we have a French guy that thinks more like Citroen than Porsche………….
@Kevin I don’t agree. If we were starting out with a clean sheet of paper to develop an all purpose CAD program you could do that.
In this case SE specialises in mechanical stuff and ID hasnt been part of their mindset.
I think the hardcore mechanical types would rebel if they had changes in their environment just like a lot of people didn’t want Realview in SW at the time.
What I was floating there was leveraging what they have and making something more atuned to ID to keep both set happy and make the task easy for SE covers.
Having tried SE I don’t think I would be happy with them just adding in some new surface tools and calling it good, it has to flow differently. It needs some extra mechanical types wouldn’t have a use for.
Rather than stuff an alternative UI and tailor made tools in there as well it would make sense to develop something marketable in its own right and in parallel at their leisure.
Their usual development team can get on with what they had planned and the ID people can follow along and reinterpret and augment as they see fit.
I don’t see what’s wrong with that. People know Alias is made for ID purposes rather than modelling plant…
No one is accusing one specialty of having no ability or interest in another. 😉
I really don’t see why Alin needs to replicate this info on the SW forum. What are they going to do with the info anyway? Nothing.
They already have 5000+ ER they aren’t going to get to. Probably they could trawl 500 more ideas from past forum discussions that they aren’t going to get to either.
Even if they do tackle things they usually end up delivering something different that is half baked.
It is apparent to most people DS don’t want to develop SW any further and the coders are otherwise employed on cloud duties so thoughts of SW getting these things are pretty much a fantasy.
Probably the most important question people have for SW presently is what are they going to substitute for the loss of Tsplines rather than what should be improved in surfacing. Let them address that first.
SW can start their own surfacing discussion if they need to or Alin could volunteer to do a search and compilation of old ideas for reconsideration if he still believes in SW.
Rather than the old guard repeating themselves in vain perhaps he should engage with the Twitter generation of users and see what they want for SWv6.
Really I think Alin is clinging to a time gone by when relations between SW and users were different and SW was going somewhere.
Sorry but when I make a contribution to this topic its not intended for the benefit of DS/SW.
After the grief people seek new beginnings. This could be a lonely year for VARS who thought life would always go on the same way. 😉
EDIT: sorry I discovered the link in #26 is from mobile here’s the desktop
http://youtu.be/vgBIeozJU2g
@Alin
Alin,
Yeah, you can post it where ever you like. I trust you’ll handle things appropriately. Sorry I’m kind of funny about promoting my own stuff.
@Dave Ault
I have a Romer arm and am in the same boat. I looked at the reverse engineering plugins like Dezignworks, but they are asking 2X more than the cost of a seat of SW for only the entry level so they can go Faro themselves.
Matt, CAD already is a monopoly ruled by a few big names. The point I was making was that they ate currently offering fragmented complex products that rely on an Eco system of bought in technologies for the so called mid range while the enterprise systems tend to be self contained systems with home grown tech.
The choice then is do they continue with this approach or do they bite the bullet and focus the product on fewer options? I’m not advocating an evil emperor but rather a savvy innovator who can take a diverse range of product, unify the interface and allow access to all the modelling goodies in the core system. My mantra is if you can’t model it you can do f all with anything else so don’t tie our hands any longer and give us the tools to create what we need.
There will still be opportunities to upscale the product to more niche areas beyond geometry creation. I just find it ironic that in the last few years we have seen visualisation move from being specialist and difficult to being easy and fun for all. We need the same for geometry now.
@matt
matt, we ran into the relax spline problem a couple of times. it’s even worse if you parametrize your edit points, build a feature with this parametrized spline and then change the parameters ‘from the outside’ by double clicking on the feature and changing the value. basically you have to go back to ‘edit sketch’ and relax the spline.
we also lost money by machining parts where we relaxed one of two identical splines on a symmetrical part after we changed some parameters but forgot to relax the other spline – resulting in an asymmetrical part.
what we do now is we parametrize splines completely..all edit points, the angles, the magnitude, everything..resulting in splines with 32974239 parameters 😀
at least that way the fucker won’t move unpredictably unless i change one of the parameters.
@Neil
But Neil that is exactly the problem. There is this prevailing attitude that people who do design are arty and people who do engineering only do prismatic modelling. That is not the case. This is the issue and what separates great products from styled products and keeps the divisions going.
Having different products for engineering and design just emphasises that. Us arty types like maths and dimensions too you know 🙂
@matt
1) Spline tool in SW need only weight control (NURBS), for the rest is a good tool
2) I don’t undertand
3) SW use the Fill Catia tool because the API that call the Fill command is more simple to integrate then Parasolid API and also because produce a surface with less node
4) The unknit tool maybe can useful for you, but never used
Trim via U & V curve NX doesn’t have and maybe Catia too. You can perform with extract iso curve and trim.
5) Do you think that automotive use this tool ?
Who said that ?
6) You need this tool ?
Good Reverse engineering software cost as a CAD software.
If I was CAD Santa for a day l’d be tempted to put out a sister program called SurfaceEdge based on the same core as SolidEdge but specifically orientated to ID users – slightly different UI presentation, tools tuned to the artistic user, a few special apps for concepts and presentation etc…and I would pitch it a bit above existing SW surfacing functionality but suited to the projects undertaken by small to medium business. Perhaps they have 5 staff – a few engineers and a ‘designer’ working on…I don’t know…custom motorbikes or something. They use SoE and a seat of SuE ( http://m.youtube.com/#/watch?v=vgBIeozJU2g ) together to great effect and everyone lives happy ever after. 😀
Matt,
Two great lists. Personally, I don’t have much need for #10 on either list, but I can certainly see where I might if my job were just a little different.
Also a great discussion from everyone. I wish SW were listening.
For what it is worth, I tend to use the control polygon, not the handles. I seem to get into less trouble with them. I think part of that is because they don’t seem to allow as extreme changes.
Jerry Steiger
@Alessandro
Allessandro, I’m going to guess that some key concepts have been left out in the translation. I really don’t care what SolidWorks does at this point. They have lost my interest. Until they show that they are interested in benefitting me, I don’t see any reason to ask them for anything.
My discomfort with the Relax Spline tool is the fact that it is necessary. If splines didn’t get kinked up, you would not need it. It is dangerous to not use it, but not everyone even knows about it, and you never know when a spline needs relaxing.
@Kevin Quigley
Yeah, I don’t get why high end modeling is associated with PLM. PLM is partially administrative, modeling is technical.
I’ve got to say I think Steve Jobs closed “ecosystem” is about as anti-customer as you can get. You have one offering – take it or leave it. I’ll take the Android model over Apple every day of the week. More options, more flexibility, less big brother, better price range.
If all CAD ever falls into the hands of a maniacal tyrant like Jobs, I’ll become a forest ranger. I think the problem is that each small CAD company tries as hard as Jobs to lock customers in, but they don’t have the resources to act as a monopoly. What you seem to be looking for is a Galactic Emperor to rule a CAD monopoly. I hope that never happens. ;o)
I’d be happy to see some product design surfacing options in a mid-range CAD system.
@Dan Staples
If someone like Santa Clause would be listening, and looking for priorities, I can give my point of view. I agree with Quigley that create and edit are separate issues, and that create is the more important of the two, partially because you have to create before you can edit, and also because one method of edit is to just revisit the create workflow, like SolidWorks does currently. I think it will work well with the current Synch Tech implementation, with create being feature/history based, and edit being either history or direct.
I think it needs to be oriented so that it doesn’t exclude technical surfacing or artistic surfacing. Some ID guys get annoyed when you call their stuff “art”, but that’s the distinction as I see it. So you need to be able to do good clean aeronautical surfaces, ship hulls, propellers, and other complex engineered shapes as well as Japanese anime figurines, shampoo bottles, and decorative scrolling floral reliefs.
Anyway, here are my priorities:
1. If you don’t have curves, you don’t have anything. So the first thing has to be good curves, in 2D and 3D. You need to be able to use combinations of entities, like surface edges, 2D curves, 2D curves, and special type curves like helix and equation driven curves.
2. Bluesurf like feature with c2 and weighting on all sides, can loft to a point from open or closed loop, can loft to a line from closed loop, some way to control internal UV flow, like draggable handle points.
3. N sided patch. SW’s Fill is from Catia. Not sure why they didn’t use the Parasolid tools.
4. Surface body tools. The Unknit (explode a face or all faces out of a solid or knit surface body) and trim by UV would be particularly satisfying, since we don’t have these now.
5. Direct edit of surface CVs.
6. Lay surface/curves over point cloud
Hi,
most request are present in hight-end CAD.
It’s possible only through R&D.
Have you present how much cost R&D ? Time to develop and time to test ?
Remember that Parasolid is not Dassault proprietary.
Why develop on a kernel that is not your ?
Sure SolidWorks can improve itself, but they have made a decision.
Decision to make an important switch.
In this time, all pass in second plane.
You have to wait and have patience.
The priority is not your request, but keep the work at this time.
About relax spline, it’s a good tool, maybe dangerous for some invalid users and an option to disable it, can be the solution, not to remove.
But you know what relax spline do ?
Relax spline improve the entire curvature of the spline.
Spline tool is better in SW then in SE.
NX has handle as SW to manipulate control point.
Spline tool in SW is very intuitive.
Here’s something I have been puzzled about for years. Why do software companies restrict advanced surface modelling to real high end megabucks packages? If I wanted to have many of the features on these lists I would need to invest maybe £20k plus around £6k annual costs (at least). That would be £26k before I even passed go – per seat.
Now whilst that sort of cost might be acceptable to a large multinational making high value products (automotive/aerospace/ships/Apple) it is harder for a SME to justify costs like that, and harder still to calculate ROI against perceived benefits.
What smaller companies resort to doing is cobbling together solutions based on lower cost systems that provide toolsets that offer as much or more functionality than the single big system. For example top end CATIA or NX (assuming you have no real need for all the PLM/PDM stuff – and I don’t know many who use that) has been replaced by SoldWorks/SolidEdge, Rhino, Tsplines, Modo etc.
Such cobbled solutions are less than ideal – multiple interfaces, translation etc, but they are a hell of a lot cheaper to implement (with much less onerous licensing as well).
I say this because I’ve just finished reading the Steve Jobs biography and I was thinking there are parallels between the closed Apple system (Mac/iTunes/App Store/iPod/iPhone/iPad/iCloud) and high end enterprise CAD/PLM systems. The other approach of cobbling together tools is more like Microsoft/Android/Linux.
The great irony of this (I think) is that Apple in the early 90s was dead on its feet because they tried to milk the professional specialist markets and ignored the mass market, only selling high cost workstations and not showing much in the way of innovation. It took the return of Jobs to realign the company and target the mass market (which, ironically, was the original target market for the Apple II and Mac).
Now, 15 years later, Apple are probably the worlds most respected brands, making a lot of money, yet providing products that more can afford (compared to the Mac II era of the late 80s and 90s).
It is not a direct comparison of course, but you can’t help but think…..what would really happen if Dassault were to offer a single modelling solution for CATIA/SolidWorks that included all the goodies and sold it for SolidWorks prices (or NX for SolidEdge prices).
We can only dream I suppose….
That is what the CAD industry needs. A Jobs like figure to come in, buy up a big name, standardise the offer on one solution (thereby streamlining interface, support, development) then sell it direct for a mass market single worldwide price. Instead, what we have is the standard industry model of costly fragmented “solutions” that not even VARs understand how to sell and support, where critical add in products can be bought up by competitors or left to wither. As an industry – this is Apple in the early 90s.
Thanks Matt. Will do.
Can I post the link on my blog too?
This sorftware exists ! It is called CATIA 🙂
It is very simple – people listen when Matt Lombard speaks. I know I do.
Plus, these are great ideas and I, for one, like to comment/discuss/brainstorm ideas. 🙂
@matt
Matt, have you tried the “Relax Handle” or “Relax all Handles” options? In the first case you get more local control.
On top of everything else said:
-A function like rough offset in catia that would also be integrated into a thicken or shell command.
-Allow self intersecting geometry.
Interesting discussion. I rarely use direct editing of spline points – I always use CVs. I watched Dan’s videos and thought “that is exactly what I do with SolidWorks” when I’m doing things like baths and basins. In my experience, direct editing of spline points leads to more errors with infections, too many points and lack of spline control. When I need a spline to touch a surface/point/line I usually create a point on the spline (not an editing point but a point) and mate this to the spline with a coincident mate, and then use something like a tangent mate from the spline to the line/surface/curve. This way we get total dimensional control.
I also tend to break splines up and use G2 continuity to join them rather than one big spline with lots of points. I find this makes things more consistent (I learned that one of an Alias trainer who trains all the automotive companies in Europe).
In terms of surfacing features there are two elements:
1. Creation of surface
2. Editing of surface
For 1. what you need is a top notch interface that allows the user to create 2D planar and freeform 3D curves with ease. None of the mainstream CAD systems are good at this. SolidWorks os good for 2D, but for 3D curves it is clunky. By far the best interface I have come across is the Ashlar-Vellum one that dates back to the 80s. For creating freeform 3D curves it is direct, simple and quick.
Allied to the curves is the ability to create surfaces from the curves, edges of existing surfaces and solids, and then control how the surface is generated. This is where the likes of SolidWorks etc do well and apps like Ashlar Vellum don’t. The biggest single innovation was using feature previews so the user can see how changing various creation parameters affect the surface. In reality I think SolidWorks Boundary surface is pretty good – as are most equivalent tools in other apps.
2. Editing. There are lots of elements to this. I did an Alias course last year as we were considering buying it (in the end we decided not to as it was the most crash prone app I have ever used). What that taught me was the huge range of evaluation tools Alias has (we were training on Surface Studio). SolidWOrks has zebras and curvature but compared to Alias and other more specialist surface modelling systems (even apps like Ashlar-Vellum Cobalt and VX) they are basic. If SolidWorks is to get to the top notch of modelling systems it has to include more evaluation tools. Specifically – more surface reflection display types, more zebra options, the ability to display surface UV isoparm curves and underlying CVs (ESSENTIAL), proper draft and undercut display. Just take a trip to Autodesk Alias Surface Studio and copy all the evaluation options in that.
Then we come to editing surfaces. Enabling display of underlying CVs allows direct editing of the CV. I’ve had this capability in Ashlar Vellum since 1998, ditto for VX and Think3. Yes you have a half baked version of this with Freeform but it is not good. The critical aspect of surface editing is to be able to make tweaks to the underlying curves, the surface CVs and the joint lines WHILE MAINTAINING CONTINUITY TO ADJACENT SURFACES. This is where Rhino is crap, and Alias is better, but where systems like CATIA/NX even SolidWorks are more consistent. An edit should be an edit – not a rebuild.
In recent years I have also been using more sub-div modelling systems like Modo. This is a completely different approach but for some types of shape it is the only way to model. This is where T Splines comes into play. The ability to create a complex junction, control edge weighting and tweak with precision is definitely the future.
But like history vs direct modelling, I think you need a system that combines both the traditional curves/surface nurbs approach and the sub-div surface push pull approach. That would be my ultimate system. That is what you have with CATIA with Imagine and Shape installed, and that is likely where Autodesk will take Inventor and Alias in the next few years.
What many users who dabble with surfacing and most CAD vendors fail to understand is that what a typical industrial designer (or technical surfacing expert) needs is precision control. Most surfacing demos you see are a pile of crap – creating totally wild surfaces and big organic shapes. Most of the use of technical surfacing is not like that. It is about defining subtle features – think a car body panel – big smooth surface with some precision detailing. Think an Apple product – the same.
One of the biggest issues for me is creating surfaces that can be edited after feedback from toolmakers. Often these tweaks require things like adjusting a freeform surface to have a minimum draft from maybe 2 to 3 degrees, or altering the surface to allow for a better design of split line. This stuff currently can take hours or days of work. If I could find a tool that enabled me to do these kinds of tweaks in minutes I would buy it. The nearest I had to that was Think3 Global Shape Modelling and Zone modelling. Great technology, bad overall package.
When splines first arrived in SW they just had an arrow on the end of the handles to control everything and it made them pretty hard to manage for practical tasks. I think I’m right in saying isolating rotation and weight via those little widgets was a suggestion of mine during the beta that subsequently appeared and that helped a fair bit maintaining your intent while manipulating the spline and later constructing the surfaces you wanted. Later they added relations and that extended the ability to define the geometry in respect of other geometry. I am not sure CP allow that.
I prefer handles to the control polygons because its a bit like drawing directly on a piece of paper and seems to fit in with the creative process better. Frequently you are tracing over a concept sketch for reference. There again I see a distinction in the way things need to be for ID use compared to mechanical work. A spline for mechanical use is probably describing something mathematically deliberate like a wing section whereas ID people are obsessed with swoop and continuity and aesthetic refinement/iteration. The trouble has always seemed to be that SW splines didn’t quite want to behave themselves and I have always attributed it to slightly dodgy coding rather than being a consequence of using handles instead of CP. I use SW2009 so I haven’t seen the end curvature issue Matt recorded but that looks to be an unacceptable behaviour from the users view.
There is another scenario Matt didn’t mention that gives trouble in that splines will frequently flip out into a big loopy tangle for no apparent reason when you work with them. This is maddening when it happens as you either have to undo and do again or delete and do again and frequently you get the same result shortly after. In my experience you even have to abandon what you wanted to do or go about it another way because it is so persistantly ignorant.
Its just one of those exasperating nonsensical flaws, like flipped mates, that they never have seemed to have tamed year on year.
Perhaps those errant little moments arise from the way the code for the handles has to be that introduces more variables to be solved and no one stable solution or obvious way of distinguishing between what is wanted and not.
If SE have a mind to improve their ID suitability I think though I would still prefer handles but minus the strange and frustrating behaviours if possible HTH
@Dan Staples
Dan, I’ve had this discussion with Mark Biasotti, who has been SW’s surfacing guy for the last several years. Mark says he prefers the CVs. Other people also use them. I tend to use the spline points. I try not to use the handles because I’m superstitious about them causing kinks. Plus, if you don’t use them symmetrically, they seem to have other adverse effects on internal continuity. I haven’t tested that out much, I just try to not use (especially internal) handles for splines. I don’t find the control polygon/CV method very intuitive. It’s a little abstract compared to manipulating a point right on the spline itself. I could work with either method, though. I see the advantages of the CVs.
Here’s a movie about using Control Polygons (CVs) in SolidWorks splines:
And another about automatic end of spline curvature, kind of annoying:
@Alin
Alin, I don’t really care to make waves on the SW forums anymore. They’re kind of on their own. If you want to post links to them there, that’s ok, but for several reasons, I don’t want to try to use the SW forums to try to divert traffic to my blog. I send people here if I have something informational, but not to promote ideas.
I guess I don’t have a good reason for not wanting to do it, I’m just squeamish about it. Sorry I can’t make more sense than that.
Just out of curiosity, why would you want me to post them on the SW forum? I should have posted them in the Top 10 submissions, but I didn’t.
@matt
Great example Matt, thanks. That’s exactly the same problem I had. When the mold maker opened the part in his version of SolidWorks, the shape changed and the mold/part weren’t as expected. Since it was a limited R&D part run, it wasn’t a serious problem, but a problem none the less.
I believe this was SolidWorks version 2010. I checked the files carefully. The difference was about .010″ in some areas along the spline. The part was about 5″ x 9″ x .040″ thick. Very stressful for me as the designer.
I don’t remember if Relax Spline was in 2010.
Devon Sowell
@matt
Got it. Very helpful. I think it was unfamiliar to me because we approach it differently (not better/worse, per se, but differently). We don’t do the handle-bars edit thing (by choice) but rather use the approach of exposing the CVs (Control Vertices/Points) directly to the user. This is kinda old school in some ways, and initially I was opposed to it, since we were doing a completely new surfacing system in 2002 and wanted to use the most modern approach possible. But when we found a way to let you edit via CVs or Edit Points (the pass through points) interchangeably, we kinda liked it better than handlebars. I guess the thing about this and kinks is, you have to work a whole lot harder to get a kink and when you do its obvious why the kink is there — because you can see where the CVs are. (Does Works have a way to toggle on the CVs and edit them? — this would probably let you fix kinks more directly than relax, which is a bit more global)
Here is a video that shows how the CVs are exposed and how you edit via the edit points or the CVs interchangeably. It also shows our local/global mode, which at the time (2002) was pretty cool and patent worthy — I’d be interested in what others do here these days. Global will preserve the shape of the curve when you make an edit, whereas local will only affect the CV you are moving — all others stay put. You can see in the video how I purposefully put in a kink but you can see it and then easily edit it out.
http://screencast.com/t/q8mrWeqP
The other cool thing about having the CVs available is we all know that if you make the first CV H/V with the endpoint then you are guaranteed tangency across that direction if you do a mirror or whatever. It makes for some very easy ways to control symmetric swoopy stuff like this:
http://screencast.com/t/llkXDvHazzQ
My point with all this is not to say woo hoo, SE has great curves stuff. I do think its quite good in many ways, but can surely be improved. You are the guys who best know how to improve it, and I am interested in rekindling the handlebars debate. As I said, we originally kicked it to the curb on purpose as having some nice “seems neat” properties, but with downsides insomuch as you are doing a lot of funky math underneath that may not be intuitive to the user (i.e. kinks). So the question is this (to all of you, not just Matt) –> Tell me more about handlebars — why do you like to edit that way (or not). In what use-case scenarios do you think tweaking the angle of the handlebars is more useful/productive than editing CVs and why?
@Dan Staples
Well, two of you have mentioned it but its not clear to me what “it” is:
“Splines would never need to be “relaxed”. When splines get kinked up, you can have 2 different states of the spline without any of the control points or handles moving. This is dangerous, and has cost me money”
Can you clarify? Maybe an Avi or something? Jing is great for that kind of thing…
As an intermediate surface modeler, I have nothing to add to the great suggestions above. I like all of them. Thanks for the post.
I’ve had this problem for years;
“Splines would never need to be “relaxed”. When splines get kinked up, you can have 2 different states of the spline without any of the control points or handles moving. This is dangerous, and has cost me money”
Devon Sowell
Along the lines of Ricks no.8 – an auto smoothing routine to tweak spline handles for the sweetest curvature combs possible with the mimimum alteration of shape.
Possibly some integrated graphics/sketching tools that can be used to generate geometry like Alias/Catia.
Hey I’m beginning to dream again…now if only SE would make a commitment to support surfacing/ID needs as well as they do for mechanical that would be very cool 😉
My self interested vote is for #10. I REALLY want to be able to use my Faro Arm directly with SE for reverse engineering. Please insert really about 10 times here.
If Solidworks surfacing actually worked we would not need this discussion. Since there are so many bugs to fix they might as well make it work nicely in the process.
1. Sketching splines should be difinitive, one spline handle does not affect any other. Spline handles should be simplified, middle spline points should have direction and two magintudes of influence. Any spline can be made proportional without respect to the current constraints. XY proportional splines would allow different scaling for two directions when used in a sweep. Circular fillets have direction and stability to never become cusps.
2. Sweeps should take a section and sweep it along a path maintaining orientation of the section plane relative to the path and maintain contact with guide curves. The resulting sections should relate to the section. If the guides are incompatible with the section and path it should stop there and do what is possible.
3. Lofts will naturally interpolate between sections, and exactly have the shape of each section. If sections have any parallel surfaces these should make a surface that is parallel between. Rounded section features should try to maintain a radius of curvature that is between that of the surrounding sections.
4. Boundary surfaces can have any number of bounding edges with good control of the boundary conditions. Guide curves will not be restricted to those with pierce constaraints.
5. Any sketch can be projected onto a surface and can be manipulated on that surface.
6. Fillets will always succeed.
7. Shell will do the sensible thing and make a shell. Tight radius corners might make a blade like internal corner.
8. A new function of smooth surface, rather like fit spline. This will simplify and smooth all contours of a surface.
9. Aerodynamic fairing. A solid body will be enveloped in a low drag shell. A tightness parameter will limit the volume of the shell. A local flow direction is specified. The resulting shapes will be rounded on the leading edges and have smoothly faired sides and a sensible aft cutoff or rounding.
10. Splines through imported points have continuous and smooth curvature.
11. Geometric features can be organized by me and grouped into folders with sub folders. Features do not own or consume curves or sketches. Parents are always ahead of chidren.
12. Surfaces are subtle and look complex but are defined by very little geometry. The file sizes relate to the underlying geometry and are thus 1200X smaller than Solidworks files. All of my projects will easily fit on a stick, with room for 100 other designers work.
No forget about posting on the SW forum we are looking to move on….SW sux under DS management. We’re interested in CAD, remember, and DS gave that up a while ago.
Besides its probably all stuff they heard before several times and did nothing about.
I take it the head SE elf would like to know what all the unhappy SW children would like next Xmas if they had a few wishes that could come true. Matt’s list is a good one although it probably fills a sack rather than a stocking, hey but then again why not be bright eyed and optimistic 😉
I would add to the list of things to fix in SW, flipping splines and tangled loft connectors – the code has never worked well enough to use in practice and wastes the users time wrestling with it. Tools must work properly and all the time.
Additions in SW- dimension spline lengths and the exact positioning of boundary and loft connectors ie. exposed variables are important for precision even if you choose not to be specific all the time.
Things to have in new software – ghosting I think is quite useful, easily accessible variables right in the 3d window, ability to chop and change selections and types without having to delete the feature and start again, more granular undo/redo, ie everything should be easy to get to and change/iterate until the user is happy, decent technical help notes explaining the basis of the tools.
I’m certain to think of more…
Matt. Would you be willing to post these lists (I believe it is not for the first time you posted such wishlists) on the SW Forum?
Excellent. I want all of them!
BTW, step 10 exists somewhat in SW’s add-in “ScanTo3D”. That being said, it would be great if the functionality you mentioned can be added to it (or to basic SW directly).
I would add the offset surface feature in step 4.