Modeling Challenge: Fades

Some things are more difficult to model than others. Some seem simple, but when you go to do them, they aren’t.

Here’s a helmet I did several years ago. The helmet wasn’t easy, but the scoops all around it were less easy. Blending the brim of the helmet was another trickfor another time, but let’s just look at those airscoops for now.

Here’s the modeling challenge for this time around. Start with a sphere, and make some scoops on it. How many different ways can you actually make work? Remember to use zebra stripes and deviation analysis to get a feel for how successful you were. Submit models to my email at matt at dezignstuff dot com. I will add screen shots of all the attempts along with a little commentary. Let’s see how many methods we can come up with.

==============================================================

Ok, the first entry is from Mark Biasotti of SolidWorks.

Mark’s entry has a surface with double curvature, and then a cut that is driven by a directional sketch. Mark’s part also uses an atomic bomb fillet in the corner, shown here in cross section.

This is a fairly simple approach, and gives a bottom of the scoop with a planar face rather than a double curvature face. It is also a solid approach, as opposed to a surfacing approach. So you can keep this in mind if you have to do a scoop as a solid rather than using more tedious surfacing commands.

One thing I want to call attention to in this model is Mark’s use of Surface Cut features. Notice that he is using planes to cut the solid model. This is maybe not a well understood function of the surface cut feature, but its very useful, and extremely stable.

Thanks, Mark!

==============================================================

Entry #2 is from Ben Paprocki. This is another solid approach.

In this part, Ben used a revolved cut to make the scoops. Revolve cuts create scoop surfaces with double curvature, but in this case the curvature is constant in both directions (although it is different in the two directions). This kind of technique works great on a spherical shape, but not on a more general complex shape where the curvature varies in both directions.

Nice model, and thanks for the submission, Ben!

==============================================================

Entry #3 is from Ben Bailey. This is a surface approach.

This is closer to the approach I would choose. Nicely done. Click the image to download Ben’s file. In this case he did the thicken basically to get the edge for the low side of the scoop. There are other ways to do this, but this certainly works. Then he made a boundary surface from the blend edge to the edge created on the offset surface. Each model submitted gets better. In this case, the main thing I would improve on would be to not make the scoop go down to a feather edge, but rather to thicken the entire outer shell, such that the scoop face would have the standard thickness as well. But this way is certainly valid. Download the SW file and check it out.

The only modeling criticism here is that there are some areas where the tangency didn’t work. See the edges on the lower side? They don’t show with a phantom line, meaning they are not within the tangent tolerance. The Deviation Analysis shot below demonstrates “how” out is out.

I would probably eliminate a split line feature, and make the entire curved section with a single boundary rather than two boundaries and a loft. Loft is said to be less accurate than boundary.

This is really picky criticism, by the way, but its the kind of stuff that keeps me up at night sometimes. How much does stuff like this matter? Depends. If you’re a perfectionist, everything matters. If you’re going to polish the mold a lot, it might be detail that just costing you a lot of money. You have to make this kind of judgment when you start working on complex models. Do you have to make it 100%? 90%? 70% How good is good enough?

Very nice job, Ben!

==============================================================

Entry #4 is from Fernando Mota. Wow. Nearly perfect. Fernando did 2 models that were both nearly perfect. A lot of work went into the first model for sure. Let’s look at the first model.

This is a fully surfaced example. The scoop face is built simply from two boundary profiles. Keeping the profiles simple is one of the best ways to get a natural looking smooth surface.

The sides of the scoop are boundaries. Again, just the way I would do it.

Fernando did a really nice trick with what appear to be fillets, but are not fillets at all. Because fillets are made from a circular section, they only give c1 transitions (abrupt change in curvature) to the faces around them. You can only get a c2 transition (smoother change in curvature) by using a spline section. SolidWorkscan only do spline section fillets if you use a face fillet with hold lines, I believe. What Fernando did was the manual equivalent of creating c2 face blends.

First, he swept a circular tube around each edge he wanted to fillet. Notice that the ends at the top, where all of the faces blend back into the main face would end in an infinite radius fillet because theoretically all of the faces are tangent. This type of thing is almost always a problem for SolidWorks. So the fact that Fernando did this manually rather than relying on the Fillet functionality was definitely the right decision.

(Another method would have been to knock out holes at the ends where the fillet would come down to a point, and then fill the hole later. This is the technique used by Mark B in the first entry, and is called the “atomic bomb” fillet, because it was originally considered to be massively overkill, but definitely fixed whatever modeling problem you were trying to avoid.)

After sweeping the tube, he used the tube to trim the surrounding surfaces, which essentially gave him an equal offset edge all the way around, and a clean edge at the top straight line transition. Then he just used boundary to fill it all in. Very clean.

Click the image to download the part.

Below is the image of the deviation analysis. Congratulations, that’s great.

==============================================================

Entry #5 is also from Fernando Mota.

This model is much simpler, and I think just as valid as the first one. It is a bit more flowing, but uses less control. Or I guess it uses the benefits of the boundary feature instead of being more direct and manual about the fillet and blend from the main surface to the scooped surface.

In this case Fernando created the scoop face, trimmed back the outer and scoop face, then just blended directly from the outer edges to the edges of the scoop, with no need for fillets. To me, having fewer faces in a model is something to try to achieve. If your goal was a more flowing surface, this would be the way to do it. If you needed to look a little more faceted, or controlled, his previous method would be better.

Again, click the image to download the SolidWorks part.

This part again shows an excellent Deviation Analysis.

Great job, Fernando!

==============================================================

Entry #6 is from Mark Rizzuto.

One of the things I’m looking for here is alternative methods, and Mark has provided that.

This is a solid technique that was performed on a flat part, and then the flat part was hit with the Flex feature to create some curvature.

Here’s what Mark had to say in his email.

Hi Matt!

Big fan, read your surfacing book and have followed your blog for a couple months now. Never submitted anything to the modeling challenges before, but this one intrigued me. After several sub-par attempts, I settled on a pretty simple approach and think it turned out ok. If I had more expertise with either the Flex or the Deform features, I probably could have turned out something pretty nice. Let me know what you think.

I add that to spur someone on. There is a technique you can use with the Deform feature that Mark mentions that might help you create a feature like this. If no one comes up with one, I might just show it myself. I’d like to see a few more entries here. These have all ben very nice models.

==============================================================

Entry #7 is from James Canney. James used the Deform feature in a way I don’t commonly see. He made a solidbody and used it with Deform, Surface Push setting. It’s kind of like denting a car door by kicking it. It’s not a very precise or controlled method, but it does give a smooth result, and it will work on any type of shape, regardless if it is single or double curvature.

Nice job with an interesting technique. Thanks for the submission, James!

==============================================================

Entry #8 is again from Ben Paprocki.

This is a bold use of the Deform feature. It’s not that successful, but it is bold.

SolidWorks has created some tools that are nearly useless for real world modeling. They are close, but not really ready for use on production models.

In this example, Ben is pushing the limits. I definitely am impressed with how good he made this look. The fact that it’s still messy is not Ben’s fault, it’s a limitation of the tool. Look at this zebra stripe display:

See how the stripes are very jagged? All of the surrounding stripes are very smooth, but the stripes in the deformed areas are jagged. This points to really uneven curvature. It looks like a dent in your car door that you tried to bang out with a small ball peen hammer. I was going to use the Curvature display, but that looks bad even on good surfaces.

Here’s how Ben created this feature:

The Curve to Curve Deform option is better than the surface push when it is used to drive existing edges. In this case it was applied right in the middle of a face. I am amazed that Ben got it to look this good, because it only gets nasty as you get very close to the face. Notice the shape options at the bottom of the propertymanager. This is what allowed a very narrow sketch line to make such a wide indentation. The way it works is that Ben told it to deform the split out face area from a sketch that was on the existing surface to a new sketch that represented where he wanted the face to go. In reality, this is a direct editing tool, because it can operate on any geometry, native or imported. Other CAD packages have tools that function somewhat like this that are called Global Shape Modeling. It manipulates existing NURBS data.

This is a cool tool, but unfortunately, it is prone to stuff like this, with the tiny little ripples in the face. It is neither very controlled nor very smooth. If SW could develop this some it might be more usable. Great example of this kind of functionality, though, Ben!

==============================================================

Entry #9 comes from Matt Wallace. This one is very interesting, a very different approach. Matt used a combination of solids and surfaces. The scoop face was a loft from an inner to an outer edge, and then he made a solid body, lofting from the scoop face to an outer face. Right, you can loft from a face to a face to create a solid.

Anyway, then this lofted body was used to indent a constant thickness solid.

Indent! I love the indent feature! This is a cool usage, and another technique we have not seen before. Cool. Thanks, Matt W!

==============================================================

Entry #10 comes from Itai Lewin. This is a surface approach, with the major feature being the Flex. Itai created the tombstone shape, and then used flex to bend it down. Then the gaps along the side were filled with a lofted surface. The Zebra stripes display of this one looks good, and what I like most about it is that it doesn’t create a split face where you want it to be smooth.

Nice job using a difficult feature to get good results from!

==============================================================

Entry #11 comes from James Canney. James used a solidsweep to indent a surface body. Excellent. I wasn’t even sure you could do that. This model has some examples of things that can go wrong even when you do everything right. These are some of those things you are just powerless against sometimes, or might lead you into crazy workarounds.

James did a nice job at everything. He even used the atomic bomb fillet technique at the corners of the fade, which are the most problematic areas on parts like this. The problem is that even though the features are symmetrical, one side splits and the other does not. This kind of thing is frustrating. You have to go back through and make sure things are exactly symmetrical, and then maybe see if you can get acceptable results by mirroring the body that didn’t split.

Ugh. That’s heartbreaking when stuff like that happens. I have no idea why, and can’t come up with a way around it except by trying a lot of different things.

Anyway, the second thing that went wrong here was with the thicken feature. Sometimes when you thicken a face, SolidWorks breaks it up into a bunch of smaller faces. This is usually because SW thinks the curvature in areas of the face is larger than the thickness of the thicken feature. That gives you stuff like this:

The most worrying things here are the little spikes. Sometimes with junk like this you can use Delete Face to get rid of all the little slivers, and just replace them with a single fill feature that’s nice and clean. If it did this in one direction, if you turn the thicken in the other direction, it might not fragment in the other direction, but there is no guarantee. The part was thickened to the outside, and the outside in that area is where the smallest radius is.

Anyway, nice part, cool techniques, and lots of stuff to look at.

==============================================================

Entry #12 is modeled by Juan Santocono of T-splines, but submitted by Matt Sederberg. Here’s what Matt had to say about this challenge:

We’ve bookmarked your blog, and can’t stay away from your contests since I think they’re one of the best sources of testing T-Splines against freeform CAD workflows. Feel no obligation to post our entry, since this wasn’t created in SolidWorks, but since we created the model we thought we might as well send it to you, and feel free to share it if you’d like.

This was done by our designer Juan Santocono using T-Splines for Rhino. You’ll notice two of the models have smooth, fillet-like edges and one has sharp creases.

The models were made creating a NURBS surface from curves, rebuilding it and converting to a T-spline, deleting some faces, merging and welding some new faces on the sides of the fade, and control point manipulation to get a smooth topology.

Hope you’re well,

Matt

==============================================================

Entry #13 is from Eric Irwin. This model was originally part of something else, as the in-context offset shows. This example builds up the scoop with surface features while the rest of the part issolid, and then uses a replace face feature to incorporate the surface body into the solid body.  Good work, and a nice looking fade. Thanks for sharing models!