Architectural models and scale prints for practices

A drawing set explains a building. A model makes people commit to it. That is why scale models still appear in planning meetings, client presentations and competition submissions long after everyone in the room has a photorealistic render on their laptop. A physical object gets picked up, turned round and argued about in a way a screen never quite manages.

The obstacle has always been cost and time. A traditional presentation model meant a model-maker, a fortnight and a four-figure invoice, which is why most schemes never got one. 3D printing changes that arithmetic. For most practices a massing model can now be printed from geometry you already have, inside a week, at a cost that sits alongside a decent set of printed drawings rather than a separate line in the fee proposal. What follows is what that actually involves: what to send us, which process suits which scale, and where printing genuinely isn't the right answer.

What practices actually order

Four things come up again and again. Massing and site models for planning submissions and pre-application meetings, where the question is how a scheme sits against its neighbours. Presentation models for client and competition work, where finish matters more than turnaround. Detail and section models at 1:50 or larger, used to argue a junction out loud rather than on paper. And physical items that aren't models at all — signage letters, sample cladding profiles, mounting brackets for an exhibition stand, replacement parts for an office model that has been dropped.

That last category is worth flagging because practices often don't think to ask. If you need twenty identical pieces — repeated balcony units across a housing-scheme model, or a run of branded objects for an open day — that is ordinary low-volume manufacturing and it prices far better per unit than one-offs do.

FDM or resin? Mostly it comes down to scale

FDM printing is the workhorse for anything with bulk. It has the larger build area, the lower cost per cubic centimetre, and it handles the things architectural models are mostly made of: solid massing blocks, stepped contour plates, terrain, baseboards, roof forms. Layer lines are visible if you go looking for them, but at 1:500 nobody is looking for them, and a fine layer height on a white filament gives a clean, matt, slightly chalky surface that reads well under presentation lighting.

SLA resin printing takes over when the detail gets smaller than a nozzle. Mullions, balustrades, tracery, cornice profiles, stair treads, 1:50 furniture — features measured in tenths of a millimetre survive in resin and disappear in filament. The trade-off is size and toughness: our resin build area tops out around 200 mm in the longest dimension, and cured resin parts are more brittle than filament ones, so a resin model wants a case rather than a rucksack. In practice a lot of jobs are hybrid — FDM base and massing, resin for the one block that has to look sharp on the presentation board.

Glazing, translucency and lit models

Glazing is the one place where printing beats the traditional acrylic-sheet approach outright, because you can print the glazing as part of the geometry instead of gluing it in behind it. Clear Cast resin prints translucent and can be polished to something close to transparent on flat faces, which suits curtain-wall panels, rooflights and glass boxes. Left unpolished it reads as frosted glass, which is often what you want.

Lit models work on the same principle in filament. Thin white walls diffuse an LED strip evenly and hide the source completely — the approach we used on a translucent book-shaped LED lamp, where the whole point was a glow with no visible hotspot. Scale that up and you have a massing model that lights its own window openings from inside, which photographs far better than a model with an anglepoise pointed at it.

Getting from Revit or Rhino to a file we can print

This is where most architectural jobs actually stall, and it is worth understanding before you export anything. Architectural models are built to be looked at, not to be watertight. Walls are single surfaces with no thickness, roofs float a millimetre above the walls they sit on, and half the model is made of components that intersect rather than join. A slicer needs a closed, manifold solid. Give it an open surface and it will either ignore it or print something unrecognisable.

Three rules cover most of it. Everything that should exist must be a solid, not a surface. Nothing can be thinner than the process can print — as a working minimum, around 1 mm of wall in filament and 0.6 mm in resin, measured at final printed scale, not at 1:1. And detail below that threshold should be deleted rather than shrunk, because it either won't appear or will appear as a smear. A 1:500 model with modelled window frames is a 1:500 model with a fuzzy facade.

If none of that is a job you want a Part 2 doing on a Friday afternoon, send us the file as it stands. Model preparation is part of our CAD service: we thicken surfaces, close the geometry, strip detail that won't resolve, split the model into printable tiles and add locating pins. We quote that as a written estimate before starting, so you know what the prep costs before you commit. We work from STL, OBJ, STEP, 3DM and SKP; IFC and RVT we can usually open but will normally ask for a mesh export as well.

A representative job

The following is representative of the work rather than a specific customer. A practice preparing a pre-application meeting for a mixed-use scheme needs a 1:500 site model, roughly 380 × 260 mm on plan. That exceeds a single build plate, so it is split into four tiles that key together with printed pins, seams placed along the two streets that cross the site so they read as roads rather than joints.

The contour base prints as one set of parts in white PLA; the proposed blocks print separately in a light grey so the new build reads against the existing context, and so the client can lift phase two off the board and talk about phase one on its own. Roughly 34 hours of machine time split across two printers, prepped and dispatched inside four working days. A single 1:200 resin print of the corner block goes alongside it for the presentation board, where the shopfront glazing needs to look like glazing. The whole thing lands in the low hundreds of pounds rather than the low thousands, which is the number that decides whether a scheme gets a model at all.

What we can't do

We are a printing workshop, not a model-making studio, and the distinction matters when you're deciding who to brief. We don't paint, spray or hand-finish. Parts come off the machine cleaned up and supports removed, but as-printed — white filament and grey resin both take paint well if you want to finish in-house, and plenty of practices prefer to. We don't do full-colour printing, so a model can be multi-part in different colours of plastic but not photographically textured. We don't cut acrylic or timber baseboards, and we don't supply trees, figures, cars or scenic dressing. We print the geometry; you dress the model.

Size is a real constraint rather than a soft one. Single filament parts run to roughly 250 mm in each direction and resin parts to around 200 mm, so anything larger is segmented and keyed. Segmenting is normal and the seams can be hidden well, but they exist, and it's better to agree where they fall than to discover them on the board. And a printed model is a communication tool: it isn't a structural, acoustic or fire test piece, and nothing about a printed plastic wall tells you anything about a real one.

How to brief a model

Four things get you a firm number quickly: the file, the scale you want, the overall size you can live with, and the date it has to exist by. If the geometry is already a clean closed mesh you can put it straight through the instant quote. If it's a working Rhino or Revit model — which it usually is — send it over with those four answers and we'll come back with a fixed price for preparation and printing, normally within a working day. If the model needs splitting or thickening we'll say so in the quote rather than discovering it halfway through.