Concept
to print

The direction stayed ours. What changed was how fast we could reach it.

Every form here was art-directed. We set the intent, judged what came back, discarded most of it and pushed the survivors further. That is the work a design studio has always done. What changed is the wait: the distance between having a judgement and seeing it tested fell from weeks to minutes, so we could afford to be far more demanding about which ideas earned the next step.

Jewellery was the right subject. Small enough to print for the price of curiosity, unforgiving enough to punish a bad decision, and at that scale a form has nowhere to hide: it is held, turned and worn against skin.

So we built the workflow ourselves rather than reading about it, and carried the results the whole distance, from concept to finished piece.

Three stages, and the tool changes at every one.

The first stage is concept generation, pushed for range rather than polish: hundreds of directions, judged on whether a shape had a reason to exist rather than on whether it looked finished. The judging is the design work, and it does not get faster. There is simply far more worth judging.

The second is the part nobody shows. A concept image is not a model, so we built our own node workflow to carry a chosen form into real 3D geometry, then rebuilt it whenever the output arrived hollow, open or too thin to survive a printer. That graph is the actual study; the jewellery is what it produced.

The third is manufacture. The pieces were printed in stainless steel, cut from their supports and finished by hand. Every stage sends something back to the one before it, and a pipeline is only worth building if it can carry that traffic both ways.

Metal 3D printing, in stainless steel.

The pieces are built straight in metal, layer by layer, which is the only reason these forms exist at all: closed lattices, undercuts and cavities that no mould could ever release. Each one leaves the machine still joined to its supports, and is cut free and finished by hand.

Speed changes which ideas you are willing to back.

When a form can be directed, modelled and printed inside one short cycle, there is no reason to defend a first idea on paper. We carried several the whole way instead of choosing between them early, and let the finished pieces settle the argument.

Every piece asked the machine for something different. Some came off the plate close to intact. Others paid for their openings in support material, and most of the finishing was getting those supports back out. None of that is visible in a concept image, and none of it can be reasoned about from a render.

That is the argument for taking a study all the way. AI moved us from idea to object faster than we had ever worked; it did not decide which object was worth having. That decision stayed with the designers, and it is the same one we make in client work, where the same distance sits between a promising concept and a product that can be made.