A set that disagrees with itself

Ask somebody who has looked at a scheme once to describe it and you will get the outline back: low, long, dark roof, big overhang. Not the brick. Not the glazing module. The edge. It is the only part of a building a person can hold in their head without effort, which makes it the first fact a set commits to and the one a reader is most likely to be silently checking in every frame afterwards. The view below commits hard. A dark folded roof of shallow intersecting gables, one fascia depth held right around, a deep overhang carried on slim posts across the whole right-hand wing, and an eaves line that runs at a single height through three changes of plan. That is a decision with consequences — because the moment you go to an aerial the roof stops being an edge and becomes a plane, and the moment you go to the far kerb the overhang is very nearly the only thing still legible. Both of those frames now have something to live up to.

A multi-material elevation is a set of decisions about edges: this plane is brick and it ends here, at this line, where the panel takes over and comes forward forty millimetres. That is the part that took three meetings. The frame below has five claddings meeting on one building — a dark red brick entrance bay running the full height, a lighter red brick base, a pale stone panel, white render above, and a charcoal panel returning the corner — and each of them has its own plane and its own edge, which is why the composition reads as deliberate rather than busy. Now consider what that costs the rest of the set. Every subsequent view is holding five materials, five boundaries and the depth relationship between them, at whatever distance it happens to be taken from. Get the brick colour a shade off between two frames and it will pass, honestly; brick moves under light and everybody knows it. Put the render where the stone was and you have not made an error of finish. You have moved a line somebody signed off.

Every other surface on a building looks broadly like itself from anywhere. Glass does not. Turn the camera thirty degrees and it goes from transparent to a sheet of sky; take it at dusk and it goes from a sheet of sky to a warm lit interior; take it flat-on under a bright overcast and it goes very close to black. All of those are correct. That is genuinely what glass does, and a set that renders it identically in six views has done something worse than inconsistency — it has lied about the most alive material on the building. So the fact that has to carry is not the appearance. It is the specification: the tint, the reflectivity, the frame colour, the mullion rhythm. In the frame below you can read the trees, the fence and a strip of the far grass in the glazing, which tells you exactly how reflective the specified glass is, and the frames are slim and dark on a regular module, and one warm timber-clad reveal turns the corner between two bays. A reader who sees clear glass in view one, a black mirror in view three and a blue-green tint in view five is not looking at a building performing under different skies. They are looking at three different orders.

Every set has a component that appears in almost every frame and gets thought about in almost none: the guard, the rail, the handrail, the newel. It is small enough to feel like a detail, repeated enough that its rhythm becomes part of the facade, and — this is the mechanical reason it drifts — it is one of the few things in an architectural scene that is genuinely expensive to carry at full resolution into a wide view, so it is a standing temptation to build a simplified stand-in for the long shots and keep the real one for the close ones. Two versions of anything is two chances to disagree. The frame below is a close read of exactly this element: a white painted timber balustrade with square balusters at a set spacing, a flat top rail, a newel at every post, repeated along two levels and turning down the stair without changing a thing. At this distance the rhythm is the architecture — take the spacing out by twenty per cent in the wide view and the building acquires a different period, which is not a subtle change, and is not the kind of thing a reader forgives as a rendering artefact.

This is the one that catches good practices, and it is nobody’s fault in particular. Interiors and exteriors run on separate tracks — separate files, separate weeks, often separate specialists — and the seam between them is a handful of elements that appear in both and therefore have to agree twice over: the size and head height of the opening, the colour and slimness of the frame, the depth of the balcony, whether the floor steps down at the threshold or runs flat through it, whether there is an upstand, what the balustrade is made of. The frame below is standing on that seam. A full-height sliding opening with a slim head sitting close under the slab, an interior floor that runs flat to the frame and continues outside in a different tile, planters standing against a glazed edge, a balcony deep enough to hold a chair and a table with room to walk past. Every one of those is also a line on the elevation. And a reader moving between the interior sheet and the exterior sheet is checking that they add up — not deliberately, not with a scale rule, but they are checking, because a person cannot look at the inside and the outside of the same opening without trying to put them together.

Look at what the ground is doing in the view below. A block-paved forecourt with a banded border, a raised kerb line holding it against the road, lamp columns at a set spacing, benches, trees at a stated maturity. That is not a texture. It is a design, at roughly the same level of decision as the elevation behind it, and it is the layer a reader spends the most time standing on — the near view stands on it, the aerial is almost entirely made of it, and the street view is judged by whether the building meets it convincingly. It is also the layer most likely to be assembled last, under time pressure, from a different reference than the one used a week earlier. Paving bond, kerb line and lamp spacing all carry. Planting maturity carries hardest of all, and it is worth being blunt about why: a semi-mature tree in one frame and a nursery stick in another is not a rendering difference. It is a five-year difference, inside a single submittal, on the exact question — screening, canopy, softening — that the planting was put there to answer.