PowerSurfacing from IntegrityWare
PowerSurfacing is another SolidWorks add-in to extend the surfacing capabilities within SolidWorks. It was brought to light after Autodesk snatched up the TSplines tool that had been getting a lot of traction in the SolidWorks universe. One of the most outstanding features of TSplines was the ability to work in mesh (subdivision modeling) and wind up in NURBS (real CAD data). This is also one of the outstanding features of PowerSurfacing.
It’s probably inevitable to compare PowerSurfacing and GeometryWorks, so let’s just get this out of the way up front. These aren’t really competing packages. GeometryWorks is really about adding several niche tools to SolidWorks surfacing. GW is traditional surfacing. PowerSurfacing represents the somewhat uncomfortable collision of the math-heavy CAD NURBS world and the art-heavy CG world.
PowerSurfacing can act as an introduction to mesh modeling for engineers and designers who have worked only in NURBS. I’ve believed for a while that the future of CAD is going to include, among other things, mesh data, and mesh techniques. It has to. NURBS modeling of complex shapes is still just too hard, and that’s the primary thing that mesh does for us. Mesh modeling works much like direct edit, in that the primary workflow is to start from a primitive, select control points, then alter. It seems like a completely alien way of working until you realize that it’s so similar to NURBS direct edit techniques.
Engineering modeling is going to get more and more organic for shape and stress optimization. The main barrier to this happening now is that you really need a specialist to create complex shapes in NURBS. Engineering keeps learning from nature, and one of the things we learn is that the most efficient shape isn’t always a straight line. The cost of manufacturing is another barrier that keeps us drawing lines and arcs instead of making more organic shapes. Molding and casting help us transcend manufacturing related shape limitations.
One of the implications of all this that you might not get right away is that putting mesh data inside a CAD tool opens the door to *.OBJ type data. *.OBJ is the file format from 3dsMax. Mesh, subdivision, point cloud are generally not engineering grade stuff, but the internet is a huge repository for free models for reference or background or rendering fodder in this format. Every wannabe graphic artist with a bootlegged copy of 3dsMax is out there making *.OBJ models for download.
Maybe even more significantly, along with *.OBJ comes 3D scan data / point cloud. Being able to work with all of this data is a huge boon for engineers and designers.
The basic idea with mesh modeling is that you tweak a mesh of points connected by lines that represent the outer surface of your part, and then the software puts a skin over the points. The game is that every mesh modeler has a different way of selecting and moving those points, and there are even differences between 3 sided meshes and 4 sided meshes. The selection and moving method in PowerSurfacing is a lot like direct editing methods. This is inherently low-resolution work, but detail is not reason you use mesh modeling. You use mesh modeling to get smooth shapes easily.
If you’re interested in getting into this side of things, you should just watch Mark Biasotti’s demo. Mark takes the time to explain why he’s doing what he’s doing, and does it from a SolidWorks point of view. The other demos tend to leave out “why” and seem like they’re explaining subd modeling to someone who already knows subd.
The development of this tool over the last several months has been pretty dramatic. Whether its correct or not I have no idea, but it would be easy to guess that the development of this tool was in direct response to Autodesk acquiring TSplines. I’m sure Dassault’s acquisition of Simpoe is an extension of the response to Autodesk further acquiring Moldflow and HSMWorks.
The interface of PowerSurfacing within SolidWorks is modern enough. It fits well with the SW interface, and is easy enough to follow. Most of the tools on the CommandManager tab are also on the RMB 
menu. I would prefer that the RMB had text labels, but that’s just me. You’ve already got the icons without the labels in the CommandManager, and learning without labels is tough. I know companies do this for a variety of reasons, not the least of which is “localization” – translation into a dozen languages to suit everyone’s needs. But really, can’t you make this available for at least English?
Another part of the interface that doesn’t thrill me is that you have to select before you can do anything. That might seem like an obvious thing, but in systems like this where you have highlighting as you mouse over something, you should be able to just click and drag some default action. In stead, you have to click to select, then click and drag an action handle. If you took the most commonly used action and made it the default, you could reduce mousing around by a significant chunk.
In general, I have had two main criticisms of subd to NURBS systems. First, that you really couldn’t plan for draft in early model development. If you were going to add draft (at the parting line) it had to be some hoky hack instead of something precise and clean. The second criticism was that these modelers were great for starting a new project, but weren’t really set up to start in the middle of a project. For example, you can make new stuff, but can’t edit existing SW models. Both of these issues appear to be addressed, or at least in progress in the 1.2 version that is available now. If you’re interested in reading about this, here is a link to a document with some of that info.
In terms of stuff I would still like to see improved, there would be a couple of things. First, get rid of this demo. Nothing against Paul, but he made this software look archaic and miserable to use. This was actually the first PowerSurfacing demo I saw, and I became very discouraged right off the bat. This is mostly because of the manual meshing. Certainly this is something that can be done automatically given a mesh size. And someone tell Paul that “intricate” does not equal “blobular”.
Next, I’d want to see saved selection sets. There might be edits you make frequently, and instead of re-selecting the points and edges, just have the selection saved and named.
Oh, and maybe the software could ship with some live demo parts. For folks who are just getting started with subd, which I would guess includes roughly 95% of the potential customers for this package, staring at an egg shaped blob and willing it into an alien head seems like a leap. Examining a real part may help keep starters from getting discouraged. Demo parts should demonstrate that subd is capable of doing something other than game controllers and other egg-shaped blobs. The main thing you’ll have to work against with selling this software is customer perception. When you think of CG methods, you think of blob-shaped characters in Wall-E, or the Incredibles, or Shrek. So I’d demo some parts with edges.
The final thing I’d like to see is to have this stuff really integrated right into a major CAD package. So thank you for developing this cool tool for SolidWorks. Now please sell it to Solid Edge. ;oP
Oh, yeah, and somebody please tell subd folks that the singular of “vertices” is “vertex”, not “verticee”. Word police today I guess.
@Ricardo Freitas
Yeah, thanks for that. That’s the kind of stuff I’m talking about. The thing is SW can do it, but it taxes all the non-creation tools, the body management tools, like knit, trim, extend, untrim, and so on. Also, 3d splines, especially spline on surface goes nuts with this stuff.
Thanks again for the perfect examples, Ricardo.
I shot some photos on a few small bits and pieces. Most things on the photos are somewhat small, going from 4 (most of them) to 9-10 inches (sample 1 and 2). I don’t know how to put images, so I just leave a link for an album containing said photos. Tomorrow I will try to scan sample 2 and 8. Any other I can take front, side and top/down photos as well as some measurements.
Anyway, here is the link:
http://imageshack.us/g/1/10161764/
Is any of this relevant enough? Or I don´t need to bother?
@matt
My family used to have a classic furniture workshop and I have access to a really extensive library of samples. I will try to find something similar to what you show in the picture. That would be completely free of property issues and I might even be able to get a scan of it, or at least a partial scan.
I will look into it sometime before the weekend and get back with some photos.
Matt,
All comment pictures have turned off for me. How do I trun them on?
Rick,
Some changes going on here. Sorry about the comment pix.
@Mark Biasotti
Mark,
You can just eyeball it. I can’t provide this data, I only have permission to share images. There is enough of this kind of work out there for someone to make a career out of it. I have a lighting fixture company and a furniture company with a fair amount of this kind of work.
@Rick McWilliams
Please don’t take my lack of comments on these subjects as I “don’t want to talk about them” but rather – yes, I’m very busy; in fact I’m visiting 6 customers in the indianapolis area right now where it is a toasty 91 deg. – but thank God that the humidity is not to bad… yet.
To answer questions above: A) No, we don’t do Class A surfacing (…yet) B) yes, we are looking to improve the surface analysis tools – and any input here or on the forums is scrutinized; but because “my” vote doesn’t count, please make sure you submit an on line enhancement request, as many of you as possible, because # of request do account for something 🙂 Matt, I will consider trying to do a PS of your capital – do you have anything in the way of a 2D print multiple view images of front, right etc. you could send me?
Mark
I did a variation on Matt’s capital using GW conic surfaces.
[img]http://www.dezignstuff.com/blog/wp-content/uploads/2013/05/capitalconics.jpg[/img]
I revolved a curve to make a vase shape. I tried to project a sketch onto this surface. Of course it can only be a sketch that does not have intersecting features that have to be connected to each other. So I made a sketch of the spines of the leaves, another sketch with 3 leaf outlines , and another sketch of 2 leaf outlines. I constructed a pair of planes for each leaf to set the angle the leaf edge encounters the vase. Each leaf is made up of two halves. I made a conic edge for the stem in two pieces the part that is tangent to the upper somewhat spherical part and the lower foot part. Finish with mirror and rotational duplicate.
This kind of part would be so much easier if SW supported a “project Sketch on a Surface feature” that would project any sketch. It would be especially good if points of the projected curve connected in the same way as the sketch.
By far! By far! 🙂 🙂 🙂
If Catia produces ‘butt cracks’ and ‘hog backs’ I think we can be confident they will be the best available, by far…. 😀
I am a little disappointed Mark didn’t have more (anything) to say about the curvature assessment itself but then he’s probably a very busy fellow and he usually gives the impression he isn’t happy to discuss stuff outside of his office. 😉
We can only try. Been trying for about 5 years now to find out about curvature. Some folks will say I’ve always been trying though 🙁
@Mark Biasotti
Mark, your input to this forum is very appreciated. It appears that SolidWorks has made real improvements to surfacing and related sketch entities like splines and conics. In some of your other posts, you have referred to CAD programs, such as Catia, that will produce class A surfaces. The inference is that SolidWorks will not produce Class A surfaces.
So, what will Catia do that SW will not, and how much does it cost to have a seat of Catia that will produce Class A surfaces? Will Catia surfaces be free of butt cracks and hogbacks??
I am concerned that surfacing requires very advanced operator ability, whether the platform is SW, Catia, or Power Surfacing, or ???. Rick McWilliams seems to be able to do amazing things with just SW and an add-in. Kind of like you did with your Poresche GT model. But it is one thing to produce a pretty model and another thing to produce a model that can be ported to CNC systems that will produce the desired part.
@Rick McWilliams
Zebras were improved in SW2012 (maps were changed from cube to sphere ) and I personally like to change the zebra reflection map to the “sky horizon” (or something like that) map which gives you a nice single horizon reflection.
Mark
Well looks like a no show from SW to discuss curvature display, still we can hope the Good Fairy might grant us a wish in a future release. Perhaps its just too late in the SW life cycle to expect further refinements and perhaps too they wouldn’t want to make SW too refined in comparison to what is to replace it. Actually it would be only the second time I am aware that something has been revisited in SW for a decent spruce up. Planes would be the first, not the most demanding of coding tasks I imagine. Once stuff gets into SW you tend to get stuck with it as it is. Its a pity SW haven’t had an ongoing program to make another pass over existing features. There are some, well a number, now that I think about it, of things that could have been done that would have really made the software sing rather than hum.
Too right, they have improved CC etc, but have not pushed development on the tools one needs to query the resulting surfaces. The edge deviation tool is one thing they have got right. How about curvature analysis OUTSIDE of the feature?
On limited occasions I’ve exported the SW model to ProE and Rhino to check surfaces.
Yeah, since people are placing some importance on the combs and stripes the display of them needs to be seen in terms of the expected accuracy and in quantified terms that people can relate to like deviation. The info needs to be seen in context of the size of the part and the intended manufacturing process. I suppose ID people might over do their attention to detail in this respect but there are situations where it is important to get it right, for instance when expensive tooling is to be made or when say the aerodynamic fairness is a prime goal. SW has supplied ever better curvature continuity, and now conics have also been included in the tool set, but in that respect perhaps the tools for accessing curvature are lagging behind in sophistication/refinement. Of course the tendancy to reduce the UI to a pleasant minimalist environment at the expense of useful technical info and intricate control doesn’t help matters. Sounds like a good wee project for someone to tackle for SW2015 to go along with the option of conic-tuned lofts and boundarys…
There always seems to be something to enhance in SW even if DS think there isn’t and social experiences are what’s missing 😉
To add to what Neil said about the display mesh and zebra mesh looking different, the same can be said for curvature display. Even with image quality/mesh AND curvature quality on high, curvature display can be really coarse; sometimes so coarse that the curvature appears to be different on each side of a mirrored body…
I think the image quality slider is too simplistic as well. How does SW derive the extent of the slider from low to high quality? Some real settings, like deviation, would be much better. EDIT. A deviation setting that is not limited like it is in SW… more like Rhino mesh settings.
I am not sure what the basis of calculating and displaying zebra stripes is currently. In SW2012, I think, it was changed possibly because Mark had read my comment on this blog that for Tsplines it was difficult to appraise the overall shape as the stripes didn’t flow very well across adjacent surface sheets. In that repect I think they improved quite a lot but it was never clear to me in the first place how well what we see in the view relates to the real world shape and if users don’t fool themselves when being a bit obsessive about their swoopy fairness. I have similar doubts about the accuracy of curvature combs. Again this is something i have asked about before but its something SW apparently dont want to talk about. I assume the zebra is derived as a quick visual representation of the variance in the normals of the triangulated mesh that is itself an approximation of the nurbs surfaces. However that doesn’t seem to be the same mesh that you set the resolution of in the display quality. Perhaps the zebra stripes user settings allow for the mesh of individual surfaces to be refined temporarily. Perhaps there should be a parameter displayed when zebra stripes are, that indicates a likely fidelity as perhaps ±0.01mm/100mm or whatever so users are aware what they see may not be what they really have. Possibly that info could be displayed when you query a point as to the curvature there. Maybe it would be helpful to display the mesh as a ghost with the stripes, not sure, or say multiply the expected accuracy at a point and end up with a progression of grays or colours within the stripes? I am sure there is some ingenious way of conveying more info than we have now. Even if the stripes cant be made more accurate, assuming they arent particularly at the moment, at least users could identify what is a real concern and what isnt. It would be nice if Mark could step out of his observer role here and comment for the benefit of users concerned to do the best work they can.
I prefer the realview factory environment to see shapes. Zebra stripes in Rhino are much more clear. Curvature colors seem to be calculated for each surface face and do not show a butt crack or hogback between surfaces. Of course SW does not tell users what these display modes really do.
@Rick McWilliams
I’ve been suspicious of the quality of the zebra stripe tool for a while now. I tend to now rely more on changing environments with Realview to evaluate the surface. This tends to show different results from the zebra stripe tool, and I think it might be showing what the surface is really doing a little bit better than zebra is. Anyone else noticed something similar or am I just imagining things?
@Mark Biasotti
Those curvature combs look excellent. I wonder why the zebra stripes show a wiggle.
I made a shape that is similar to the middle section of the hand unit and it did not show wiggles. It was not easy in SW to get a shape that has similar character. It is rather bent in two directions. The best tool seemed to be GW multiblend. A feature that I know even less about than the GW blend.
It looks like ps is giving you some nice shape control.
I see a modeling challenge. With the leafy capitol. I will give it a try.
@Mark Biasotti
Mark, I’d like to see some flowery furniture design. That is the stuff that most makes me pull my hair. Plus, I’ve got an active project that I’m constrained to SW2009 on.
Something like this:
There are at least 3 different examples of people modeling the game controller. It gets a little old.
Matt, you mentioned that you’d like to see more “real-world” examples (not game controllers and blobs) come out of PS. Do you have any good ideas of what type of products that you’d like to see for PS accomplishing?
Here ya go Rick:
[img]http://www.dezignstuff.com/blog/wp-content/uploads/2013/05/gamecontrollercurvature.png[/img]
The nice thing about this is while in PS, I have symmetry on and I can dynamically pull on sub-D on right side side and the other side (mirrored side left) updates automatically. This is very useful for tweaking the curvature across the mirror plane – abet PS does not have curvature combs – although I have met with them and we agree that it would be a useful future enhancement.
@Rick… could you post a image or images of the “hogback” and “butt crack” you talk about?
Thank you.
@Mark Biasotti
The smooth butt crack is shown by the wiggle in the zebra stripes in the plane of symmetry. I think one side is a hogback. If you can show curvature combs the curvature will show a reversal. Sometimes this kind of feature is sexy.
I think that my mannequin, Mr Bill, would be very well defined by power surfacing methods rather than boundary surfaces.
@Rick McWilliams
Rick – can you please identify the “smooth butt crack” on the game controller? If it’s in there – I intentionally put it there.
Mark
Mark,
Do all SW complex surfaces need to have hogbacks and butt cracks? Your shape is smooth maybe G2 but it has a smooth butt crack. I am sensitive to this type of shape defect. It is probably more conspicuous in the curvature combs.
It’s a tough sell right now, if looking at the long term big picture. We have it here at the studio I do some work from and it great that it’s native inside of SW. It was a huge deal break that it wasn’t able to work with native SW data. I’m glad to see that they are making this happen in the updated version….
@Jeff: It was big discussion in the studio about software, combination of…etc. From a pricing standpoint consider that one has SW already and you’re looking for “organic” modeling, and better rendering capabilities. And since we all know that breaking links sucks because every time data is exported/imported from one program to the next there’s just a loss in one thing or another there’s just no one easy solution.
At that same price point for Power surfacing you get Modo which does SubD WAY modeling WAY better than P-Surf and then throw in the kitchen sink right along with it. Kicking out a model from Modo to SW as a NURBS need the plug in, but there’s full control over HOW the surface edge comes into SW which is a huge plus.
This is true to a point. Power Surfacing will open obj files and will open them from modo. Once opened they can be edited in Power Surfacing. This is also true for FBX files. But…………I find there are limitations in the complexity of the obj/fbx files that can be opened/edited. If its a complex mesh model it either won’t open or is so “heavy” when opened its difficult to deal with. I know this is being improved and worked on.
You could model in modo and use a Luxology product http://www.luxology.com/store/Power_SubD-NURBS/ to go back and fourth from SolidWorks to modo. This converter was also created by NPower.
In general though I think you’re better off using Power Surfacing because it allows you to remain in one application and therefore is a better workflow. My opinion.
@Troy
Sub-D modeling is not the be-all to end-all when it comes to creating complex product designs. I say that intentionally because it’s absolutely wonderful for character design but the challenge for the product design space is that most modern (and traditionally) Industrial Design is mostly analytical forms with some complexity (non-analytical) as part of of that. Some designs will clearly be much easier to execute using traditional solid modeling and some designs will be easier using Sub-D. then there are a whole lot of designs where it will be a clever mixture of both. This is where integration to solid modeler makes it generally more useful.
Additionally, Sub-d in some ways is harder to get the knack of than curve surface modeling because in many ways you have to be MORE aware of topology and what can and cannot be accomplish because of the quadrilateral approach that is so wants to adhere to; it is definitely a different approach and something that you learn and acquire over time.
Mark
All boundary connections are G2 and extraordinary points are G1.
[img]http://www.dezignstuff.com/blog/wp-content/uploads/2013/05/gamecontroller.jpg[/img]
[img]http://www.dezignstuff.com/blog/wp-content/uploads/2013/05/gamecontroller2.jpg[/img]
@Jeff Mowry
I haven’t tried this yet, but I believe you should be able to model anything in Modo and bring it in with PowerSurfacing using the .Obj importer that it adds. If you look at the options, PowerSurfacing gives you “convert to sub-D” quality options (low, medium, fine, very fine) for Obj import. I assume this means any mesh obj file from anywhere will be converted into an editable PowerSurface in SolidWorks. So in theory, yes, model in Modo, open in SolidWorks and have complete edit-ability with PowerSurfacing.
Considering Creo offers Freestyle which is a sub-d in the base package, I can only hope Solidworks partners with Integrity like they did with Simpoe to offer this as an official extension. That most Solidworks users won’t care I can sympathize with, so I’m not arguing they need to offer it in the base package like Creo.
Afterall PowerSurfacing seems to be somewhat more extensive than Freestyle anyway.
p.s. To add on the singular of vertices is vertex. also the plural of vertex is vertices; not vertexes. I’ve heard that one a lot, too!
I wonder if a similar work-flow exists between modo and SolidWorks? Does anyone know more about this, and to what degree edits can be made between the two?
By the way, great to see this sort of thing. Thanks for the recent posts on alternative means of creating geometry!
I absolutely agree that the future of CAD needed Mesh sculpting tools and other similar approaches to modeling that all work seamlessly together. This is the kind of functionality I have been waiting for as long as I have been doing CAD. It’s funny to me that it had always been considered impossible for so long, and no one could do it. Then all of a sudden everyone can do it? What’s up with that?
@Neil
This kind of tool is definitely more of an “art”. It takes a lot of practice and completely new ways of thinking about how to approach modeling a shape. It’s very easy for beginners at this to get a very lumpy shapes. I’ve said a few times now that it is not suitable in a lot of situations. The key is to try and figure out what situations it is good for and which situations it should not be used in. This is something we are currently in the process of trying to figure out with our one license we have purchased so far.
The ability to convert a SolidWorks surface into a Sub-D for sculpting is a critical feature I feel, and one of the reasons we decided to buy this. Right now we are figuring out how to model an overall form the standard way, and then use Sub-D in only one area to freely sculpt a tricky part like a grip or handle area on a bottle. The real trick is to figure out how to blend the final Sub-D area back into the rest of the model. There are a whole bunch of new tricks and techniques you have to learn that are very different from any strategy you might have developed in the past for modeling things. At some point this will all lead to whole new set of “Best Practices” to be followed.
There was a question about how accurately the Sub-D is converted to Nurbs, and the software nicely comes with a few different options for controlling that. I’m pretty satisfied so far with it’s conversion capability.
One additional feature I would really love to see is quarter symmetry. There are quite a lot of shapes I create that I only need to model one quarter of and then mirror it twice to get the full shape. I think they have said they are working on that already though. And that is the other reason we bought this thing…the guys developing this have been very open and responsive to questions and are really developing this very quickly and doing a pretty good job at it too. It’s refreshing to see this kind of effort being put in to something again. It gives me a lot of confidence that they are really serious about this. And that means a lot in a purchasing decision.
I took the time to watch Mark’s demo. I think PowerSurfacing suffers more than Tsplines in how the shape changes as you are working. It seemed to me Mark was chasing the geometry around a bit, but he got there. One critical thing that seems to be missing is curvature combs. If they are available I didn’t see it in the UI. It isn’t clear how faithful the conversion to nurbs is either. I wonder what tolerance operates for the best fit and if you can specify it. Although Mark succeeded in making a volume similar to his sketches I don’t think it would have been necessarily very fair since each minor vertex misalignment, movement or a subdivision would have introduced a subtle hump/hollow/wave or knock-on effect. This is something that is difficult to handle in Tsplines as well. In my experience subd is very useful for some things but can be a bit frustrating for others particularly where you have a need for precisely engineered shapes. Perhaps it would be better to break down the mission into a few surfaces that can be mutually trimmed and blended in SW rather than try to achieve everything in one hit in a well massaged blob.