Subtractive, Additive & Formative

Subtractive processes start with solid stock and cut material away until the part remains. The family includes CNC milling and routing (a spinning tool carves a block or sheet), laser cutting (a focused beam vaporises a line through sheet material), and waterjet (a high-pressure jet of water and abrasive slices almost any material cold). Their great strengths are precision, surface quality and speed on the right material - a laser can cut clean 6 mm plywood in seconds; a mill holds tolerances of a few hundredths of a millimetre.

Their limits all come from the same fact: a physical tool has to reach the material. A milling cutter cannot machine an undercut or a fully enclosed cavity, because the tool cannot get behind the surface - this is the defining geometry constraint of the family. Cutting also produces waste: the material removed becomes chips or offcuts, and the cut itself has width. That width is the kerf - typically around 0.1-0.5 mm on a laser, wider on a router or waterjet - and it must be accounted for or slotted joints come out loose. Sheet processes fight waste with nesting, packing parts tightly on the stock. Subtractive is usually the right first thought for flat sheet work, precise metal or wood parts, and moulds - which is exactly why most real fabrication is subtractive, not printed.

Additive processes build a part by adding material, almost always in layers, straight from a 3D model. FDM / FFF melts a plastic filament and lays it down bead by bead - cheap, robust, the desktop default. SLA / DLP cures liquid resin with light, giving fine detail and smooth surfaces for jewellery-grade parts and moulds. SLS fuses nylon powder with a laser, and because the surrounding powder supports the part, it makes complex geometry with no separate supports. At the other end of scale, concrete and clay printing extrude construction-grade paste to build walls and formwork.

The superpower of additive is geometry: it makes internal channels, lattices and organic shapes that no tool could reach, and it wastes little material. The costs are equally characteristic. Layer building is slow, and it imposes two constraints every designer must respect. First, overhangs beyond a certain angle need throwaway support structures, which cost material and leave marks. Second, parts are anisotropic - a typical FDM print is markedly weaker in the Z direction, so it tends to split between layers under load, and you must orient the part with that in mind. A worked example: print a hook lying flat and it may snap along the layers at a fraction of the load it survives when the layers run along the pull. Additive is not a synonym for digital fabrication; it is the family you reach for when the geometry genuinely cannot be reached any other way.

Formative processes keep the same volume of material and simply reshape it. The family includes bending and folding (sheet metal on a press brake, acrylic on a line bender), thermoforming (heating a plastic sheet and pulling it over a mould with vacuum), moulding and casting (pouring or pressing material into a cavity - resin, plaster, concrete, injection-moulded plastic), and vacuum-forming signage and trays. Digital fabrication rarely does the forming directly; instead it makes the tooling - the mould, jig, buck or die - by CNC or 3D printing, and formative then reproduces that form quickly and cheaply.

Formative shines at repetition and material efficiency: once a mould exists, each part is fast and near-waste-free, which is why the whole mass-production world is formative. But it carries two honest limits. It needs tooling, which is an upfront cost and time, so formative rewards quantity and punishes true one-offs. And materials fight back - a bent metal sheet springs back by a few degrees when released, so you overbend to compensate; a thermoformed sheet thins over deep draws; a casting shrinks as it cures. The digitally fabricated mould is where the designer's control lives, but the material has the final say on the form. Recognising a job as formative early tells you the real deliverable is often not the part at all - it is the mould.

This is also why formative and the other families are so often paired. A designer CNC-mills or 3D-prints a positive or a mould once, then casts, presses or vacuum-forms a run of identical parts from it - marrying the geometric freedom of a digital tool to the speed and repeatability of forming. A concrete example from architecture: milling a reusable formwork mould, then casting dozens of matching facade panels. The digital step happens once, up front, and the formative step multiplies it cheaply - a pattern worth recognising because it turns an expensive bespoke shape into an affordable series.

No family makes everything in one piece, so a fourth activity stitches parts together: joining and assembly. It spans mechanical fasteners (bolts, screws, dowels), integral joints designed into the geometry (finger joints, mortise-and-tenon, laser-cut slots and tabs, snap fits, the flexible living hinge cut into a single sheet), adhesives, and welding or solvent bonding. The digital-fabrication move is to design the joint into the file so parts self-locate and self-align - a slotted rib that only fits one way is worth more than a page of assembly instructions.

Assembly is also where robotics increasingly enters the field. A six-axis robot arm with the right end-effector can place, glue, weld, nail or stack components with a reach and repeatability no other machine offers - stacking bricks to a computed pattern, assembling timber frames, weaving fibre. That is the whole subject of Modules 6 and 7. For now, the point is that real projects almost never sit in one family: a typical build is laser-cut ribs (subtractive) that slot into a thermoformed skin (formative) with 3D-printed connectors (additive), all bolted or robot-assembled (joining). Fluency is knowing how the families combine, and designing the joins so the assembly is obvious, strong and forgiving.

With the families clear, choosing between them becomes a short interrogation of the job. Geometry first: is the shape essentially flat sheet, or a solid with reachable surfaces? That points to subtractive. Does it have internal cavities, lattices or undercuts no tool can reach? That points to additive. Is it a form a sheet or a cast can take? That points to formative. Then quantity: one or a few unique parts favour subtractive or additive, where the setup is the file; hundreds of identical parts favour formative, where a mould amortises across the run.

After that, weigh material (a laser loves acrylic and ply but not thick metal; a waterjet cuts almost anything; FDM is limited to thermoplastics), scale (desktop printer versus a gantry that prints a wall), tolerance and finish (milling and SLA are fine; FDM is coarser), and waste - a real design criterion in a circular-economy world, where subtractive offcuts and additive supports are both losses to minimise. There is rarely a single right answer, and the best makers combine families deliberately. The discipline is to ask these questions before modelling, because the family you choose reshapes the geometry you should draw - a part destined for a 3-axis mill must avoid undercuts, and a part destined for FDM must respect overhang angles. Choose the family first; let it inform the form.