Choosing between filament, resin and PolyJet

How to pick a process from the requirement rather than from habit.

Choosing between filament, resin and PolyJet comes down to naming what the part actually has to do first, structural strength, fine cosmetic detail, or several materials combined in one piece, and letting that requirement point at the process, rather than reaching for whichever machine happened to print the last job successfully.

Three machines sat within arm's reach of each other in that lab, a filament printer, a resin printer and shared access to a PolyJet, and the choice between them stopped being a matter of preference the moment a part actually had to hold together under load, or look presentable in a meeting, or do both at once, because each machine answered a different one of those demands well and the other two poorly.

Introduction and overview

Fused filament fabrication melts a plastic filament and extrudes it layer by layer, producing tough, genuinely structural parts cheaply, at the cost of a visibly ridged surface and strength that depends heavily on which direction the load is applied relative to the layers. Stereolithography cures a liquid resin with light instead, producing a smooth, highly detailed surface with strength that is much more uniform in every direction, at the cost of a part that is generally more brittle and needs careful post-processing before it is truly finished. Material jetting, the process behind PolyJet, jets tiny droplets of photopolymer and cures each layer as it goes, which lets a single print combine several materials, rigid and rubber-like, in one job at the finest detail of the three, at the cost of being the slowest and by far the most expensive way to make any given part.

The marker-pen-and-watercolour comparison

A thick marker sketches a bold shape quickly and holds up to being handled roughly afterwards, exactly the trade a filament print makes, fast and tough at the cost of a visibly coarse line. A fine technical pen trades that speed and robustness for a precise, clean line that reads well up close, exactly what a resin print buys, fine detail and a smooth finish at the cost of a nib, or a printed feature, that chips more easily under real handling. A full watercolour set does something neither pen can, blending several colours and textures within a single piece of work, which is what a PolyJet print does with materials rather than colours, at the cost of far more setup time and a far higher price for the privilege.

Why the requirement should choose the process, not the reverse

Defaulting to whichever machine is most familiar, rather than asking what the part specifically needs, quietly optimises for the operator's comfort instead of the part's actual performance, and the mismatch usually only shows up once the part is already in someone's hands failing to do its job. A bracket that needs to survive being dropped belongs on the filament printer regardless of how much nicer a resin version would look sitting on a desk, and a small connector that needs to seal precisely against another part belongs on the resin printer regardless of how quickly the filament machine could have produced a rougher version of the same shape.

The habit of defaulting to one familiar machine tends to compound over time rather than staying harmless, since every part successfully made on the familiar machine reinforces the sense that it is the right tool in general, right up until a part with a genuinely different requirement arrives and fails in a way that looks, at first, like a one-off surprise rather than the predictable result of a requirement that was never actually checked against the process chosen for it.

The number that matters here

A part with several genuinely different requirements at once, rigid in one region and flexible in another, can sometimes only be produced as a single piece by a PolyJet's ability to jet more than one material within the same print, a capability neither filament nor resin printing has any equivalent for regardless of how the part is redesigned around their limits. That capability alone justifies PolyJet's far higher cost per part for a genuine minority of designs, while for the great majority of parts, which need either toughness or fine detail but not both materials combined in one piece, filament or resin alone answers the requirement at a fraction of the cost.

What this changes in practice

Naming the actual requirement before choosing a machine, structural load, cosmetic finish, or multi-material behaviour, turns a habit-driven choice into a deliberate one, and the habit worth building is asking what the part must survive or display before asking which printer is free or familiar. Doing this consistently across a year of parts meant fewer reprints caused by picking the wrong process the first time, since a part sent to the correct machine from the outset rarely needed to be sent to a different one afterwards once its first attempt failed to hold up. It also made the later articles in this set easier to write, since almost every specific failure covered from here on, a part that delaminated, a part that kept curing after it left the machine, a print that could not have been made without a support material that dissolves away, traces back to exactly this same first question of what the part actually needed before any machine was chosen for it.

Where this stops being true

None of this means the cheapest or most familiar machine is ever the wrong default for a genuinely low-stakes part, a quick jig, a one-off fixture that will be thrown away within the week, where the cost of guessing wrong is trivial and the time spent reasoning carefully about requirements would cost more than simply printing it and finding out. The framework earns its keep specifically on parts that matter enough to be used, shown, or relied upon afterwards, and applying it with equal weight to every single throwaway bracket printed in a busy lab would slow the lab down for no real benefit in return.

My key error with this

My default was the filament printer for anything that was not obviously demanding, on the straightforward grounds that it costs almost nothing to run and I already know how it behaves, and I applied that default far past the point where it made sense. What the reasoning left out was my own time, since a part that comes back needing a second attempt, or that is dimensionally close enough to argue with, costs hours that are worth considerably more than the difference in consumable price even on a machine where the material is expensive. The second omission was subtler and took longer to notice, because filament does not keep well, so a wide stock of materials held for the occasions they are needed is a stock quietly expiring, and the only way to actually have that breadth available is to use it often enough to keep it turning over. What replaced the belief is that the cheap process is not automatically the responsible one, and that a default which never gets re-examined stops being a decision at all.