Plastic is fantastic. High impact polystyrene and ABS are wonderful materials for making miniatures from. They are lightweight, flexible and elastic enough to be resistant to light bending, but hard enough to be easily cut and shaped with hand tools. It’s also cheap. If you buy a box of hard plastic miniatures today, the box probably cost more in terms of materials than the plastic inside of it. So why aren’t all models plastic, and what’s the catch?

Lets start with the most basic explanation of how you get from an idea to a miniature.

A sculptor produces a sculpt from something like epoxy putty (think: greenstuff, magicsculpt, kneadite), clay or plaster, maybe incorporating metal parts or parts of other models. Or possibly they sculpted it digitally and they’ve 3D printed it and meticulously sanded and finished the model to remove all the layer lines* and other artifacts of the printing process.

They would like more than one copy of it, so they embed their single model (possibly cut into parts) into silicone rubber and cut it open to produce a silicone mould. Here they pour a two-part resin mixture into the mould to produce an initial copy, a resin master. This mould is usually only ever intended to produce a small number of copies, which are then used to either repeat the process and make a production mould in the same way, to make final resin copies; or to then make a vulcanised rubber mould which produces metal parts.

A brief digression on resin vs. plastic. There’s already someone reading this thinking “aren’t both HIPS and ABS resins, and isn’t the kind of resin they’re talking about just a thermoset plastic?” Yes, and no. Most plastics are resins, and in this context all the resins we’re talking about are plastics. But in common parlance, plastic means hard thermoformed polymers (stuff you melt and force into a mould) and resin means hard thermoset polymers (stuff you mix and pour into a mould).

The resins used for making miniatures are typically more brittle than hard plastic, but the fact they cure at roughly room temperature** means you can use low temperature silicones to make moulds which are cheap. The flexible nature of the moulds means that you don’t have to worry about undercuts as much, which allows a resin miniature to have more detail.

An undercut is where part of the model being moulded protrudes outward into the mould sideways, like a rivet on the side of a tank, or the gap between a gun and the chest of the soldier carrying it. If your moulding material is flexible, you can flex the mould out of the way, or if the material is flexible, you can sometimes flex the material out of the mould. But as we’ll discuss later, if neither is flexible, your only choice is to rigidly avoid undercuts.

Depending on the resin, you can also have big thick chunks of resin allowing for single piece large castings for big monsters or big tanks, reducing complexity, or you can easily cast large hollow pieces through a technique called rotocasting. 

Metal on the other hand is hard, bendable, and generally cheap. Until the early 2000s when RTV silicones became cheap and easily accessible, for making a lot of cheap models you’d generally turn to a vulcanised rubber mould. You take two donuts of raw rubber, place your resin masters between them, and place the whole thing in a vulcaniser (a big metal tin that keeps it under pressure) and cook it in an oven for many hours until the rubber turns hard. The rubber, being set by heat, can withstand low temperature metals such as pewter and other casting alloys. You then load your new vulcanised rubber moulds (after some post processing) into a machine called a spincaster (no relation to the rotocaster) and pour molten metal into it. The metal cools quickly and the spinning forces the metal into all the nooks and crannies. If you’re old enough to have routinely purchased metal models, you might remember the little whispy bits that were often on metal models: these were the air channels to let the air come out of the mould as the metal was forced in, and depending on how good the caster was, you might get more or less metal in there.

The process of making metal miniatures still allows for some undercuts, but fewer than resin, and you can’t have any one part be too thick or it will take too long to cool. Additionally, the whole mould can’t be too thick, or you can’t easily vulcanise it without a much more powerful oven and much more pressure, so single piece metal models tend to be quite flat or be made of many parts (each of which is flat).

So far, we’ve got resin and metal and they’ve got their own trade offs:

• It’s cheaper to make one silicone mould than it is to make a vulcanised rubber mould, but the vulcanised rubber mould lasts for more castings

• Resin is cheaper than metal

• Metal offcuts and failed castings can just be remelted, where resin offcuts and failed castings are scrap and must be disposed of***

• It’s faster to produce a large run of metal models than it is resin models as metal cools faster than resin cures and you can cast many more models in a single metal mould than you traditionally can in a single silicone mould

• Resin holds detail better than metal, and silicone moulds hold their detail better until they break, where vulcanised metal moulds degrade over time more gradually

• You can have larger resin models than single part metal models

• It’s easier to glue resin together than metal

• It’s easier to paint resin than metal

• Resin is lighter than metal

• Your humble author prefers resin over metal

With that out of the way, lets actually talk plastic. Lets rewind and consider our humble starting point. I am not going to go into detail into the pre-CAD processes (maybe for a future post) so lets assume that either the sculpt started digitally, or the physical original sculpt got 3D scanned and we have a digital copy.

Step one is the model needs to be cut and keyed. Plastic injection moulding has a number of major limitations that we need to keep in mind, and so we need to cut the model into different parts so that every part obeys these rules.

First up, undercuts. A plastic injection tool (not a mould, why is it a tool and not a mould? Complex historical reasons and also to make you, the reader, look like a rube if you say the wrong thing in front of a jackass) is a solid chunk of steel or aluminium into which the negative of our model is going to be carved. If there’s any area where the metal of the tool would overhang, this is an undercut in the model and we wouldn’t be able to pull the model straight out of the tool. So they’ve got to go.

Sometimes we can cut a part into multiple smaller parts, other times we can rotate a part to reduce the amount of undercut, but often we’re left with a little bit of an undercut, and so we have to just fill it in. If you look at the shoulder pads of a space marine with the little nubbins bonding studs on them, the studs on a plastic version all point straight out and appear to cast a plastic shadow back towards the curve of the shoulder pad. This is to prevent an undercut. If you look at the resin version, those nubs are all perfect hemispheres. This is why on the new MK IV space marine kit, the shoulder pad comes in two halves, to reduce the “shadow” each nub casts.

Next up are draft angles. As previously mentioned, the plastic needs to slide out of the tool easily, and that means minimising friction. If a model includes a surface that is exactly perpendicular to the tool, as the model is removed from the tool that surface would scrape against the side of the tool, causing friction, which is bad. So no surface is allowed to be perpendicular to the tool, and every surface has to be a little bit slanted. If you’ve ever bought a model with an integral puddle base and wondered why the bottom of the puddle has a little crest to it, that’s why. This isn’t usually a problem on most models, except on thinks like tanks. So again, the sponsons on a Leman Russ are modelled at 45 degrees and in several parts, as to prevent any perpendicular surfaces. The other place where this comes into play is cylinders. The side of a cylinder approaches a vertical surface, which would be bad, and so you end up with more of a slightly pointy oval if you want to do a cylinder in one part. It’s not so noticeable on something small like a spear or staff, but it’s why large tanks and other cylindrical parts are often done in three or four segments.

But what about a nice big part? Nope. Like with our earlier metal example, ideally we want our plastic parts to all be a single uniform thickness so that all the plastic cools uniformly, or at least not too thick, because if a thick part cools too quickly it can shrink and pull plastic away from other parts of the model. You can see the physical effect of plastic cooling on some large, flat parts like a rhino chassis. The rainbows and visual ripples you can see in the plastic as you hold it up in the light are the visual signs of the cooling plastic forming stress into the final part.

So a large, thick piece has to be broken into two (or more) hollow pieces of a roughly uniform thickness. This is why a space marine torso is two pieces. You could see the consequences of not doing this on the bottom of the old Cadian stormtrooper torsos, where the flat waist often bowed inwards a bit. This is from where the plastic shrinks back.

Additionally, our parts can’t be too thick, because otherwise our tooling becomes too thick and much more expensive. Single blocks, called billets, of aluminium or steel aren’t cheap, and tools have other features in them like cooling channels (to cool the plastic faster) and ejector pins (to automatically remove the model from the tool) have to be machined into them. Most injection moulding machines can also only accept tools of a specific thickness, so there’s often a hard limit even before the cost is considered. Beyond that, if we’ve got a part that’s highly concave, think like the top plate of a rhino, if one side of the tool is highly indented the other side has to rise up beyond the plane of the parting line (the line where the two halves meet) to fill it. Which means we might have to start with a much larger brick of metal and remove a large amount of it.

Finally, we need to talk ab out the parting line and other fine details. If you pick up a GW sprue with a trapezoidal runner, the parting line is almost always along the plane of the larger of the two flat sides of the trapezoid and runs across the entire sprue. This is, as we’ve said, where the two halves of the tool meet. Some stuff, like a flat base, might all be on one side of the line, whilst other bits like arms or heads or guns will cross the line. The line is almost always flat because that’s easy to make, and easy means cheap. If the tool isn’t kept perfectly closed as the plastic is forced in, a tiny amount of plastic might creep between the plates. If it’s just a small, raised ridge, that’s a mould line. Unfortunately, mould lines are just a fact of life, although as many thousands or tens of thousands of parts are removed from the tool, the edge can wear away a little and cause the mould line to become bigger or more pronounced. Eventually, more plastic can seep in producing a larger area of flat plastic around a part. This is called flashing and is usually a sign a tool needs to be replaced.

Just as the edges can get damaged, any area with very fine detail (like pointy bits or anywhere where the draft angles aren’t sufficient) can wear over time and become damaged, reducing detail. Ensuring that you don’t have very sharp pointy details in your design facing into the tool is important, it’s partly why the pointy MkVI helmets are modelled sideways instead of facing outwards.

Fun side fact: if you look at the runner on a GW sprue and you see a line of little raised dots near the design date, each dot is a sign that the tool has been repaired. If there’s been some damage, you can mill out that area, weld in new metal and cut it back to replace the detail. It’s not quite as good as a new tool, but it’s cheaper and faster.

So we’ve got our limitations and we’ve cut out model into parts. It’s useful to add keys in at this point, the blocks and nubbins and their receiving recesses that help you identify which parts go with which and hold them together whilst the glue cures. Now we’ve got to link them together.

It is possible to do this automatically, but it’s usually**** done by a talented person called a tooling engineer, who will take a sculpt and turn it into parts that can actually be made. Ideally, the person sculpting the part has knowledge of how it’s going to be made and has taken that into account. When this goes well, you get some absolute magic with models positioned and posed to allow for maximum detail with minimal parts, or assemblies that go together like magic.

When it goes wrong you get malifaux, and the horror of the seven part head with top hat.

Before I describe putting parts on sprues, I have some bad news. You might want to sit down for this.

The big plastic thing that goes in a box that has all the parts on it isn’t a sprue. That’s a runner. The bit that connects the runner to the model is called a gate. The only part that is a sprue is where the plastic is injected into the tool, which is sometimes visible as a cylinder in the middle of the frame of runners.

Now that’s out of the way let’s forget that and call it a sprue. Also, just to make things even more confusing, in resin casting the sprue is often called the gate, and the runner is the sprue, but gates are either sprue or gate, depending, and sometimes you still have actual runners.

None of the nomenclature makes sense.

The plastic has to get into the model, and for that we have the sprue. It moves plastic from the injection point, through the runners (the round or trapezoidal framework bits), into the gates and finally into the part (called a mould cavity). These have to be balanced to ensure that plastic ends up filling each part at roughly the same time, otherwise it can put too much pressure on the mould and blow it open, which would be bad. But also, all of the plastic you inject into a tool exerts pressure onto the tool halves, which the injection machine has to counteract.

The big modern injection moulding machines used by the likes of GW exert tens, if not hundreds of tonnes of pressure on to the tool halves to keep it shut. Improvements in clamping pressure, along with improvements in terms of design of the models themselves, is one of the big reasons that if you look at a plastic kit from the 2000s and compare it to the one released in the last few years, the old ones are so incredibly more spacious. They just didn’t have the clamping pressure to put as much plastic in as they do now.

So now we’ve got a perfect set of sprues and parts sitting in CAD, which is wonderful. Except for the fact we need a chunk of steel or aluminium to put into the machine. The whole thing gets passed over to a machinist, who will program a CNC machine to cut it out from a billet of metal using either an engraving setup using a rotary milling machine which uses endmill bits of varying sizes (like drill bits but they cut sideways instead of just down), or using a technique called Electric Discharge Machining, or EDM. (Yes, even the professionals make the joke you just made. I once attended a demonstration of a benchtop EDM generator where the manufacturer made the joke. It will never get old.) The finer the detail, the longer it takes, and the more brittle the metal can get and as previously mentioned, the more likely the tool is to break.

You also have to consider (especially for engraving setups) clearance. The milling machine head has to be able to get the cutting tool into all the places where you want to cut detail, which also pushes tools to be as shallow as possible to reduce this issue.

So finally, finally we’ve got an injection tool you can load up and pop parts out of. But you might be wondering “what about the single part cabs common in some historical and scale model kits for trucks” or “what about those fancy multicolour parts that gundams sometimes have?”. Well this is already 3200 words, I’ve got to stop at some point. That can go in part two.

So next time you pick up a kit and wonder “why was this made of several parts” or “why is there this weird plastic shadow behind this kneepad”, maybe now you’ll have a better idea as to why. And also why they can’t just take the plastic titanicus warhound and scale it up 4x in cad and call it a day.

The machine would explode. And you’d make the tooling engineer sad.

Don’t make the tooling engineer sad.

* You might be surprised to learn that often, layer lines get left in many places on even expensive models. Modelmakers are only human, and because making resin model kits isn’t a super high margin job, it’s often not worth getting them out of nooks and crannies. And silicone does an excellent job reproducing EVERY surface detail, so it might not have been visible on the original master, but is on the resin copy.

** Most modelling resins heat as they cure in an exothermic reaction, and it’s this heat that damages moulds, which means moulds fail. If you’ve ever purchased a resin model kit and found a bright chunk of silicone stuck in a crevice, this is why.

*** Ground up resin can be used as filler for future resin castings, but it’s not a major use.

**** To my knowledge GW does it mostly by hand, but companies like Wyrd do it largely automatically. This is why malifaux models come out with nine trillion parts.