Introduction: Gaming Dice Made From Folded, Waterjet Cut Steel

About: A kinetic sculptor known as Fish. He is currently making a slow, terrifying transition from computer professional to full-time artist.

Over the last few months I've been developing a construction technique for my entry in a design competition. I can't share the actual design itself, sadly, but I can share the process. So I did the whole thing again, this time with a couple polyhedral dice models.

First, a tessellated shape is created in a 3D design tool such as Blender. This is exported and opened in Pepakura, which takes 3D models and "unfolds" them so they can be printed and cut out as papercraft. Instead of paper, however, we're going to get it waterjet cut out of 14 gauge steel sheet -- including the score lines to get it to bend into shape properly. Then we'll weld it together, grind it up all pretty like, and do a surface treatment.

Step 1: 3D Design

I did my designs in Blender, though any 3D design tool could be used as long as it can export STL files. Because we're working with steel sheet, the shape needs to be made entirely of flat surfaces -- it should be tessellated like something in an early 3D game.

The final object will be slightly larger than the shape you designed. Following this process, we basically wrap the design in steel, so the final product ends up being larger by twice the thickness of the metal you end up using. On more complex designs, this can be important to remember, particularly if you have to 3D print jigs to help with assembly.

Once you are satisfied with the design, export it as an STL and import it into Pepakura Designer. (The free Pepakura Viewer won't work, I'm afraid.) Unfold it, turn off tabs, and export the result as a DXF. There are a lot of guides for Pepakura, so I won't get into the details here. If your design has to be split into smaller pieces, you'll probably want to make sure they are symmetrical. which Pepakura won't do by default. Sometimes Pepakura will add cut lines in weird places, due to the model having faces that are even slightly non-planar. I usually find it easier to fix that in the DXF afterwards with a tool such as LibreCAD, rather than fight with Blender and Pepakura.

I've attached all the files used for the dice as an example, both the original objects in Blender and the final, cleaned up DXF files.

Some general guidelines I've learned after many rounds of testing:

  • If at all possible, make sure that only 3 planes meet at each vertex (point). This makes assembly far easier, as there is only one way to fold them together. With 4 or more, this isn't true.
  • Minimize long, narrow faces. (I.e., highly acute triangles.) They're hard to bend into shape because you can't get a good grip with pliers on the very end of the tip where it is the narrowest.
  • We can bend a piece down to any angle we want, but we can only bend it up by about 45 degrees without extra work. Try to minimize bends like that in your design. (See the next section for more discussion of this restriction.)
  • Complex designs will most likely have to broken into multiple sections when unfolded. Think carefully about how they will be assembled to choose where to make the breaks. Those seams are likely to be wider and more obvious, so try to hide them if possible. This is another good way to deal with highly acute creases which are greater than 45 degrees. If you put the join between two sections down there, it's a lot harder to see.
  • Also for very complex pieces, getting the angles of the bends exactly right is critical, as any errors build up and result in huge gaps very quickly. I found the best way to solve this was to 3D print a version of the final shape, and use it to test each angle as you bend it. You can also do something similar to create jigs that fit on the outside, with press-fit magnets holding different sub-assemblies in the correct alignment for welding.

Step 2: Waterjet Cutting

Unless you have access to a waterjet cutter, you'll need to outsource this next step. Waterjet cutting is the process of cutting a material (pretty much any material) with a very thin stream of high pressure water. Very, very high pressure. Controlled by a computer (CNC), it can do precise cuts through several inches of steel. Usually when I use it, I need a large number of repetitive parts made for a project, or I'm working with very thick materials beyond the scope of my shop, like my welding table top. But it's also useful for very detailed work like the Pepakura output, which would be exceedingly difficult to cut accurately enough by hand. I've found having a good relationship with a local shop very helpful when working on techniques like this... though not always all that good for my credit card!

So, for waterjet, we're doing something a bit weird here. We're not just cutting out a shape, we also need to "score" the junction between faces where we'll be bending. Without this, we'd never get the kind of sharp, crisp bend that we need, not even using a good sheet metal brake. The Pepakura DXF output will color these junctions blue and red for you. For normal papercraft use, this would distinguish "mountain" folds from "valley" folds, that is, those folded down vs. those folded up. (The example d6 shown here only has blue, because it has no concavities. It's all "mountain" folds.) We can only waterjet cut from a single side, though, so we'll score all of them in the same way.

When I first tried this, I did perforations along these lines, like on a stamp. This was both needlessly expensive, requiring the waterjet to start and stop many times, and resulted in ugly bends. Luckily, my shop suggested a better way to do it. Get both blue and red lines etched instead of cut. For my provider, that means running at 50 inches per minute along those lines, but you'll need to talk with yours to figure out what works with their equipment and the steel stock you're using. We want to cut through most of the steel, but not all, providing the perfect score along which to bend it.

This works great for all "mountain" folds. It also works for "valley" folds, but with a limited range. After about 45 degrees the edges of the etched channel touch, and you can't bend it any farther. The best way to deal with this is to limit those angles in your design, or to choose them as splits between sub-assemblies. Being down in a sharp crease hides the join nicely. If you absolutely can't do that, you could widen the etched slot yourself using a Dremel-style tool. But since this would be labor intensive and likely to be irregular and ugly, I'd avoid it if at all possible.

Step 3: Assembly and Welding

As long as you've been careful to make sure all your faces meet in groups of 3, assembling your piece is a simple matter of bending the faces together until they just touch at the edge. One of the pictures shows this in progress with the d12. If you haven't, you'll have to spend a lot more time finessing the angles to be just right. I found it useful to 3D print a version of the design, so I had something to test bends against. But no matter what, you'll probably need to wiggle it around a bit at the end to get everything sufficiently closed up.

Personally, I think the raw fold line are quite pretty. On my original piece, which was open on the bottom, I was able to weld it together inside where it wouldn't show. This preserved the fold lines, which ended up being a strong element of the overall look. The dice, however, are completely closed, so this wasn't possible. I had no choice but to weld on the outside. I wanted a consistent look, so I welded along all the seams, even those that didn't need it.

This is a bit tricky to weld, since the steel sheet is so thin. You'll need to dial your voltage and feed rate way down so you don't burn through immediately. Get some scraps and practice first! I'm a passable welder, and while my results were solid, they certainly weren't pretty. Thankfully our species invented angle grinders.

Step 4: Grinding and Polishing

The next step was grinding. Lots and lots of grinding.

First I went over all the edges with a basic grinding disk. Clamping was increasingly difficult as I moved from d6 to d12 to d4. I wish I had some clever solution for the d4 I could share, but I really just pinched it at the base in the vise and worked on one fraction of a face at a time. Very annoying! If I was doing this regularly, I'd make the number holes on the d4 a bit bigger, so some kind of T bolt could be inserted, locking them down from below. But this worked, you just need to be patient.

Grinding revealed several places where I hadn't built up enough metal while welding, leaving divots and cavities. This was particularly a problem at the corners, where I had been worried about burning through. After finding all these places, I went back with the welder and built up a bit more metal in those places. Which meant even more grinding after that was done, of course.

Rough grinding is just that -- rough. It doesn't leave a very nice surface. So I switched out the grinding wheel for an 80 grit flap sanding disc. I inserted the protective jaw covers on my bench vise at this point, since I didn't want to mar the surfaces after I had just got them all nice and shiny.

The results looked pretty good. (Though a round of 120+ grit wouldn't have hurt.) The edges on the d12 came out particularly nice and crisp! But I knew all that raw steel would rust up really easily, and I wanted a less blinding effect anyway. Time for some toxic chemicals!

Step 5: Surface Treatment

I decided to use a basic gun bluing solution. Bluing is a generic term that covers many different processes for converting a thin surface layer of steel into Fe3O4. This is the good kind of rust. It is darker and provides some amount of protection against the bad kind of rust, Fe2O3. It's also kind of pretty! Before modern alloys and coatings, it was commonly used on guns to protect them. (Hence the name.) These days it's mostly used to be decorative. There are many different ways to achieve it, most of which are annoying and/or toxic. I went with the easiest, which is to buy a little bottle of bluing solution from a hardware store.

Safety note: Do not drink bluing solution. Do not get it in your eyes. Wear eye protection. Wear gloves. Don't be dumb.

After applying the bluing I scrubbed the pieces thoroughly with a wire brush and rinsed them off. This was followed by burnishing with a scrubbing sponge, and then a coat of clear lacquer.

I'm quite happy with the end result. I'm not sure I'd want to play with these, but they sure are pretty. (If I do say so myself.) This was a particularly pleasing project for me, because I made my first set of gaming dice back in 1989 by hand using a similar papercraft process. It's always nice to revisit old projects and see how much you've improved!