Some practical and non-practical thoughts on tool-paths.
Posted by Liav Koren on March 23rd, 2009
[Note: I hope that even if you know nothing about computer numerically controlled machining you'll find this comprehensible and/or interesting. At least it should be interesting if you're interested in making things. My friend/colleague Greg Smith recently wrote a primer on rapid prototyping in an art context. Means of Production appeared on Rhizome (which is shamefully devoid of writing on digital fabrication) and Greg's Linear Notes appear on his excellent blog, Serial Consign.]
"Hello world." Some of my recent roughing and finishing tests on the Romaxx cnc miller.
When you’re working with computer controlled machine tools, creating and finessing tool-paths becomes a big deal. The work-flow is generally something like this: you have a Widget Concept which gets represented in CAD (a 3d model, or some sort of digi friendly representation). From there the CAD model is moved to a CAM system which chews on your model and spits out toolpaths, which get fed to a controller that eats tool paths and spits out instructions to your tool. Your tool then spits out mistakes and, occasionally, widgets.
Let’s consider tool paths and a subtractive process–say, a cnc mill. You’ll usually want to rough out the form first using the biggest and clumsiest endmill amenable. After the roughing pass is finished, you’ll do one or more finishing passes with smaller bits. As you continue to do finishing passes, the trace of the tooling generally becomes subtler and subtler, but at the same time, eradicating the tool traces becomes more and more of a losing proposition: your endmills keep getting smaller, which makes your tool paths longer and more complicated and the probability of something getting fucked up just keeps rising, as does the time involved.
Leonardo DiCaprio as Howard Hughes. Hughes was obsessed with, among other things, getting rid of all traces of rivets in his planes. Scorsese seems to think it wasn't just about aerodynamics...
Of course, pretty much everything around us is a product of machining, but in general the traces of that machining are hidden. You might occasionally see the subtlest seam where the parts of a mold came together, but certainly no trace of the process that made the mold. At some point this became a formal elements of modernism and industrial design. Even in every-day life we now have a certain aversion to surfaces that have traces (ipod covers?). Gropius once called industrial civilization a “slipcover society”. He was probably referring to the use of architectural facades as a mask, but the observation is more broadly valid. Obviously, in some situations we like seeing surfaces that look like they’ve been kicked around. And clearly one aesthetic hasn’t supplanted another–it doesn’t really eve make sense to think of them as being in competition, anymore. We’re simply noting two different approaches to materiality. (Think Cloud Gate vs. Torqued Ellipse.) In general, this is an interesting topic that I’ll have to return to later.
The deeply textured surfaces of the Unité d'Habitation literally are traces of how it was made: timbers placed together made up the form-work, and left their texture as a trace. Compare the influence of the Unité to that of the Villa Savoye.
For the moment, though, the point is that the iterative process of finishing paths, and the very nature of CNC machining and work-flow gives you a huge amount of control over the traces your tool will be leaving on your workpiece. Of course, you can obliterate those traces if you have the patience and some sandpaper, but a potentially more interesting approach is to think of these traces as an integral patterning that comes out of the hardware, software(s), and your own control and orchestration of these systems. This can be a particularly powerful idea when you also start to fold in other techniques such as laminating your stock piece [1] together. Mould making also offers a world of operations and constraints that are particularly interesting.
A CNC milled mould is used as a formwork for concrete panels, by Ply Architects. Again, a sort of "hello world" level project, but still interesting. www.plyarch.com
Pinstripe. Close-up of a custom door by Williamson and Williamson Architects. The effect is produced by laminating Baltic Birch plywood on an angle. www.williamsonwilliamson.com
Even in 3D printing where you’re adding material, tool paths can leave a trace. No messy roughing passes are involved, but even on high resolution industrial printers you can often see a subtle grain or stratification in the finished piece. At the very least, thoughtful designers consider this directional element when prototyping items, although they may have to give more precedence to issues like overhangs (which 3d printers tend not to like) and stability. Projects like RepRap achieve a much less precise level of tolerance (although, at .10 mm positioning resolution, it’s still pretty respectable), and it’s at this level that machining traces really start to make themselves felt–RepRap’d components are just some of the weirdest fucking objects that I have ever seen. From some angles their sparse infill, just-slightly wonky geometry and off-white colour makes them seem organic, even bone-like. From other angles the crisp, machine-ruled rigor of their forms dominates. When you come in close, you notice small, odd deformities and flaws, places where molten ABS was pulled across a gap, or where two layers briefly delaminated. This level of imprecision is much more interesting than professionally printed objects.
Recent, good quality RepRap'd parts. Courtesy of Wade Bortz.
While RepRap hardware has gotten to the point where you can treat it as a complex Lego Set (in the sense the modules have been resolved to the point where you can conceptually just snap them together. Cf. Lego Rap), it’s also still an open R&D project and getting involved means that sooner or later you’re going to become entangled with it. While commercial 3D printers position themselves are hermetic black boxes, RepRap is like subtractive machining in the way it telescopes the production process, and really ropes you into its systems. It literally wears its openness on its frame, cladding itself in the hand-soldered circuit boards that control it.
RepRap vs commerical 3D printer. One of these things is not like the other...
It’s in this messiness that we see another way computer controlled machining challenges notions of craft. In a 2006 interview by Régine Debatty, Douglas Edric Stanley muses:
[we are entering a regime where] modularity reigns, where objects are seized by constantly shifting rules and conditions, and programmable machines create not only a new aesthetic, but even ask the question anew, “what is aesthetics?”
Often we talk about “interactivity”…But over time I have come to realize that interactivity is only the tip of the iceberg, and that the idea of building endless gadgets that go BOING! in the dark when you step on them is absolute vanity. At the same time I loathe the contemporary art world’s smug distain for these same gadgets, their incapacity to see the emerging field within these often simple, almost childlike objects and installations. So I’m trying to evolve the gadget without throwing out the charm of what takes it beyond its pure gadget status.
Stanley goes on to speculate about an evolutionary pathway for tools and devices: reactive -> automatic -> interactive -> instrument -> platform.
Site model study by Liav Koren & Janice M Lee.
Footnotes:
[1] The stock piece is the uncut, raw material you begin with.
ps – screw you, wordpress. Screw you.
