From Facade to Midsole: Why Architecture's Parametric Tools Fit Objects Better Than Buildings
Updated: Oct 6
Start with a scheduling detail that says more than it seems to. On October 21, the first afternoon of Rhino World 2026 in Castelldefels, just outside Barcelona, the guest-speaker session includes a talk on the central towers of the Sagrada Família. The next afternoon, the same auditorium hands the floor to Volkswagen Design. A basilica, then a car company, and one piece of software running through both. If you work in either world, read that agenda as a map of where your tools are going.
We're at the start of Q4, when studios lock in next year's budgets, software seats and hires. So it's a good moment for a plain question. Why are the computational design tools that architecture made famous, Rhino and its visual-scripting environment Grasshopper, turning up on product designers' desks? And what happens to the wall between buildings and objects when both sides share the same scripts?
Our answer, up front: this isn't architecture colonising product design. It's parametric thinking coming home, and the object may suit it better than the building ever did.
What the tool actually is
If you've only ever modelled directly, here's the shift. In Rhino you push geometry around. In Grasshopper you build the recipe: a graph of components where parameters flow through operations and come out as geometry. Move a slider and everything downstream redraws. David Rutten, who wrote it at Robert McNeel & Associates, has said it was developed for Rhino customers "as a way to automate tasks without the need to write textual code." Its first public name, Explicit History, makes the same point: rather than Rhino recording your history while you work, you define that history yourself.
The origin story matters for this crossover. According to Daniel Davis's history of parametric design, architects got their first visual-scripting language when Robert Aish began beta testing Generative Components at Bentley Systems in 2003. McNeel tried and failed to license it, assigned Rutten to build an alternative, and released Explicit History in 2007. Grasshopper has shipped inside Rhino since version 6, as TU Delft's industrial design textbook notes.
Now for the twist most architects forget: parametric modelling didn't start in architecture. Davis names Parametric Technology Corporation's Pro/ENGINEER, released in 1988, as the first commercially successful parametric modelling software, and it was a mechanical engineering tool. Its founder, Samuel Geisberg, told Industry Week in 1993 that "the cost of making design changes ought to be as close to zero as possible." Architecture borrowed that logic. Gehry Partners brought in Rick Smith, a CATIA specialist from aerospace, to realise the Barcelona Fish and the Guggenheim Bilbao. So the arc runs from product engineering to architecture and back to the product studio. Grasshopper is the return ticket.
How it's crossing over
The evidence is in the projects. Back in 2012, New Balance's Running Innovation Studio and Sports Research Lab worked with the practice Radlab on 3D-printed running midsoles. The lab captured each runner's stride on a force plate fitted with an array of 100 pressure sensors. Radlab read that data, stored as Excel files, straight into Grasshopper and tested four strategies (vector fields, Voronoi cells, webs and particles) for turning pressure maps into sole geometry tuned to a single runner. The business case was pure product logic: injection moulds made a run of, say, fifty pairs too expensive, and printing removes the mould.
Look at who does this work, too. Nervous System, founded in 2007 by Jessica Rosenkrantz and Jesse Louis-Rosenberg, brought backgrounds in architecture, design, maths and biology to jewellery. For their Cell Cycle line they skipped commercial software altogether and wrote their own in Processing, generating roughly a hundred variations in a couple of days before picking eight bracelets and four rings for production. Their reaction-diffusion tools later ended up in New Balance shoes. What travels is the method, whatever the software is called.
By now it's ordinary studio infrastructure. The Czech practice Polednopithecus runs a single Rhino-and-Grasshopper toolkit across facade optimisation for architects and, for industrial clients, parametric automotive grilles, CNC-milled furniture textures and lattice structures and wearables. It uses the same scripts for curvature analysis and surface-continuity checks. At Volkswagen, the Rhino World programme says Grasshopper already handles internal tool development and production-oriented workflows, and the talk sets out to show how it can "move beyond parametric modeling and become a platform for design tools, manufacturing methods, and scalable automotive design workflows."
Teaching is following the same path. TU Delft's Industrial Design Engineering faculty pitched its open Grasshopper textbook at product scale, noting that most existing material serves architecture. Theirs centres on ergonomics, digital fabrication and designs generated to fit one person.
The legitimate criticism
None of this makes the tool neutral. Three objections deserve a real hearing.
First, sameness. Architecture has already been through this. Patrik Schumacher's "parametricism", launched as a manifesto in 2008, drew years of criticism for producing placeless, copy-paste buildings. One essay points to a project that copied Zaha Hadid's Wangjing SOHO, began construction at the same time and was set to finish first. Product design has its own version: the Voronoi lattice used as surface texture. Even Nervous System say they held back from using reaction-diffusion more because it had been "so heavily tread by other computational artists."
Second, the handoff. Products have to be engineered, simulated and toleranced, and much of that work still happens in feature-based CAD. One McNeel forum thread shows the friction. A cellular structure built in Grasshopper arrived in SolidWorks as open surfaces and couldn't go into simulation until the pipe ends were capped to make closed solids. It's a small fix, but this is exactly the kind of thing that eats a week.
Third, maintainability. Davis was already writing about Grasshopper "spaghetti" in 2010: definitions so tangled that nobody can edit, test or reuse them. A studio that builds a product line on undocumented graphs has built on sand. McNeel seems to agree. Its Grasshopper 2 beta, released in the Rhino beta this July, is a ground-up rewrite. It attaches metadata to every item, spreads its solver across processor cores, lets you cancel a running solution and loops natively. It also tells users, plainly, to keep production work in Grasshopper 1 for now.
The TDW house view
Here's where we land. The sharpest critique of parametric architecture turns out to be the best argument for parametric products. In "What's wrong with parametricism", Richard Coyne argued that parametric design flourishes in facades and roofs rather than floor plans because a building's skin has constraints you can actually write down, without the tangle of use, culture, politics and human habitation. Now think at the object scale. A midsole, a ring, a grille, a lampshade: each is a bounded problem with measurable inputs like pressure, fit, airflow, load and the size of the print bed. Coyne's own opening example was a chair.
So a tool that got over-sold as an architectural style works better as a product method. Buildings gave Grasshopper its reputation, and objects may give it discipline. The wall between the two is coming down through process, not looks. The same graph can rationalise a facade panel and a car grille because both are surfaces governed by rules.
What this means for you right now
If you're an architect, your scripting is portable. Practices like Polednopithecus already sell the same computational logic across scales. Learn how products get toleranced and handed to engineering, and you become useful in rooms you've never been in.
If you're a product designer, start with data, not form. Radlab's midsole work began with a pressure map, not a pattern. If a pattern can't justify itself in terms of load, fit or fabrication, it's decoration.
Whichever side you're on, treat definitions like code: name your inputs and outputs, group the logic into functions and document it. Rhino 8 helps here. It added a new script editor for Python 3 and C# in both Rhino and Grasshopper, plus Grasshopper components for blocks, annotations, user text and object attributes, so attributed data flows into drawings and downstream tools. Its ShrinkWrap command, built to wrap geometry in a watertight mesh for 3D printing, closes one more gap between a graph and a part.
The long arc
Parametric thinking was born in mechanical engineering, grew up on curved buildings, and is now back at the scale of things you hold in your hand. The tools didn't change disciplines so much as the problems finally caught up with them. On October 22, when Volkswagen takes the Rhino World stage the day after the Sagrada Família towers, watch the slides about tolerances and production rather than the ones about curves. That's where the wall between buildings and objects actually comes down.
Comments