What a building loses when it becomes a ring

Every ring I make starts as a building, and every one of them loses something on the way down.

That loss is the interesting part, so that's what this is about.

The problem with making a building small

A building stands up because of its size. Span, load, the weight of the thing pressing down on itself. The structure exists because the scale demands it. Shrink it to twenty millimetres and the forces that justified the form are gone. What's left is a shape that used to have a reason.

So for each piece I have to pick one idea and let the rest go.

Yure: the geometry of something that was going to move

In 2015 a timber pavilion stood in the Jardin des Tuileries in Paris. Twelve metres of stacked wood, built from thin sticks of a single section, stacked and twisted. No screws, no bolts, no glue. Traditional Japanese carpentry joints, and the sticks held each other up.

The whole point of that structure is that it moves. Japanese timber framing never treated movement as a fault, because a structure that resists an earthquake outright is a structure that breaks.

The ring is cast solid. It cannot move. It is a single piece of silver and nothing in it gives.

That's the loss, and it isn't recoverable. What survives is the arrangement, not the behaviour. I decided that was honest enough to keep, but I won't pretend the ring does what the pavilion did.

Engawa: three thousand of something becomes none

An engawa is the strip of floor running along the outside of a Japanese house, outside the paper walls but still under the eaves. Neither in the house nor out of it.

The canopy at the Centro de Arte Moderna in Lisbon runs about 107 metres and roughly 15 metres wide, dropping to around two metres at the garden edge and rising to about ten at the building. Its upper face is shingled with some 3,000 overlapping white azulejos. Underneath it's lined in ash.

At ring scale you cannot have 3,000 of anything. Repetition that reads as texture across a hundred metres reads as noise across twenty millimetres, and then as nothing at all.

So the ring keeps the condition rather than the count. A band that reads as an edge rather than a boundary. The tiles are gone. The in-between is what I was after anyway.

Passage: the interval, not the elements

The bamboo passage I took this one from used strips fifteen millimetres by three, more than ten thousand of them. You walk between two ranks of verticals and the light arrives in strips, at a rhythm set by the spacing.

The building isn't the thing you look at. It's the thing you pass through, and the interval is the whole design.

That one scaled better than most, because an interval is a ratio and ratios travel. The ring is silver where the bamboo was and nothing where the gaps were.

Then the metal gets a vote

Everything is modelled in Rhino and Grasshopper, printed in wax, and cast in sterling silver here in Tokyo. The middle of that sentence is where most of the real work happens.

A definition that resolves cleanly on screen does not automatically cast. There is a floor on section thickness below which molten silver will not fill a channel. It chills before it arrives and you get a piece with a limb missing. Openwork that reads beautifully in the viewport can trap investment, so the material you're trying to remove afterwards has nowhere to go.

In practice my definitions are shaped as much by what the casting will tolerate as by the form I was after. A parameter that produces a gorgeous model and an unmakeable object is a parameter set wrong.

What actually happens is that the constraint improves the result. Being forced to thicken a strand until it will fill makes you decide which strands matter. A model that can be anything tends to become fussy. A model that has to survive a mould gets argued down to its idea.

Why I'd rather publish the failures

Studios publish renders. Renders are the easy part. The viewport never refuses anything.

What's actually useful to anyone working this way is the pile of wax prints that didn't make it, and the reason each one didn't. That's the information nobody shares, and it's the information I wanted when I started.

So if you work in Grasshopper and you're heading toward something physical: the screen will tell you the model is finished long before it is. The metal is the reviewer whose opinion counts.

About

I'm Aldi. I worked at Kengo Kuma & Associates before starting this, which is where the habit of looking at buildings structurally comes from. Everything here is designed and made by me in Tokyo. The pieces are my own work and are not affiliated with any architect or practice named above.

The full collection is at carpediemsculpture.com.