Ctrl + Couture: Issey Miyake’s Style Signature, From the Archive to the Algorithm
What does a computer see in thirty years of Miyake? This project uses image clustering to map out seasonal aesthetics and creative pivots.
I was walking the streets of SoHo with a friend when the Issey Miyake store caught my eye. I wouldn’t describe this as an average retail experience, it was Issey Miyake. Pleats suspended mid-air, silhouettes folding in on themselves like origami with an attitude. It wasn’t just fashion. It was design, it was engineering, it was movement frozen in time.
I’ve admired Issey Miyake for some time. The legacy, the silhouettes, the movement. But nothing prepared me for how it felt to wear it. Trying on Miyake was like stepping into the fourth dimension. The garments moved like ideas. Every fold and curve felt intentional, experimental, alive. It wasn’t just beautiful. It reminded me why I fell in love with data in the first place. Not because it was technical, but because it offered new ways of seeing.
Miyake redefined what clothes could do. His designs were always reaching toward the future, so using machine learning to explore them didn’t feel like a stretch. It felt like speaking his language. I scraped runway archives, processed hundreds of looks, and used computer vision to identify the five most representative outfits from each season. It became a conversation between two kinds of innovation: one in fabric and one in code. My goal was to trace the evolution of his aesthetic through raw image data and see what patterns might emerge.
Inputs: Issey Miyake
In 1970 the world of fashion would be forever changed as Issey Miyake launched Miyake Design studio in Tokyo. This studio blended tradition with innovation, shaped by a background that bridged avant-garde graphic design in Japan and classical couture techniques in Paris. He became known for integrating cutting-edge textile innovation with a minimalist Japanese philosophy, created garments that were as functional as they were sculptural.
In the late 1980s, Miyake introduced a heat-set pleating technique that would come to define his most iconic work — folds that held their shape, but never lost their lightness. His Pleats Please line made fabric feel futuristic, long before “wearable tech” was even a phrase. Later came A-POC (A Piece of Cloth), a single tube of fabric designed to be cut, shaped, and styled without waste, merging sustainability with design flexibility. Miyake’s work constantly blurred the line between machine and maker, between design and movement. The clothes he created moved so effortlessly with the human form, adapted to the changing world, and redefined what fashion could be.
Turning Style into Signal
How do you teach a machine to see fashion?
Unfortunately, machines don’t have the same taste for fashion as we do — they just don’t see a garment as garment. They see pixels, and those pixels need cleaning up.
I selected eight Spring/Summer collections spanning between 1995 and 2025, each season chosen to reflect a distinct phase in Miyake’s visual evolution. But across these three decades, runway photography has changed — and so has the data quality.
In the 1990s, many archive images were scanned from prints or uploaded in early digital formats, tinted with blue shadows or blown-out whites. In more recent seasons, crisp backstage lighting and 4K lenses flood the frame. Beautiful for humans, chaotic for machines.
But if I wanted the model to analyze silhouettes, not spotlights, I had to clean up this data. So I stripped each image down to its essentials. Backgrounds were removed, contrasts normalized, and color drained to grayscale. The goal wasn’t aesthetic: it was algorithmic. The more noise that was removed, the clearer and cleaner the signal was.
From there, I ran each image through a deep learning model trained to recognize visual features, not trends or textiles, but the architecture of an outfit. What the algorithm made was a numeric fingerprint for every look, think of it as a set of coordinates in a high-dimensional “style space”. The closer two looks sit in that space, the more visually similar they are. And vice-versa, the further two looks sit in this space, the more dissimilar they are.
I grouped the looks by season, found each year’s core style, and selected the five outfits that clustered closest to that core style. These weren’t the loudest looks — they were the most representative. They were what a machine thinks Miyake looked like in a given year.
Seasonal Style Index
Five looks that best define the structure of the season
From 1995 to 2025, the model traced Miyake’s evolving design language through shape alone — each season mapped not by color or fabric, but by form.
These looks weren’t chosen for spectacle. They were selected for structure: posture, pattern, and proportion. A machine’s quiet read on what defined each season.
SS1995 – Sculpted Volume
Under Issey Miyake
These looks reflect Miyake’s own hand: engineered pleats, bold architectural cuts, and forms that felt both futuristic and grounded in craft. It was a season about shape as structure.
SS1998 – Fluid Minimalism
Under Naoki Takizawa
Miyake’s longtime protégé brought a softer hand. The silhouettes here are relaxed, meditative—tailoring gives way to drape. The machine picks up the quiet precision of a designer settling into a new rhythm.
SS2010 – Graphic Geometry
Under Dai Fujiwara
Fujiwara reintroduced pattern with control. Garments became graphic experiments, and silhouette was used as a canvas for engineered movement. This season’s core style sits at the edge of function and art.
SS2012 – Deconstructed Uniformity
Under Yoshiyuki Miyamae
Textile innovation took center stage. Miyamae’s Steam Stretch technique reshaped the role of structure entirely. Silhouettes here play with asymmetry and density—they feel modular, technical, alive.
SS2020 – Movement as Philosophy
Under Satoshi Kondo
Kondo’s debut centered “ma”—the Japanese concept of negative space. Clothes moved with air and breath. The five closest looks are fluid, suspended, almost weightless.
SS2025 – Structured Ease
Under Satoshi Kondo
In the model’s reading of the present, strength re-emerges. Silhouettes feel solid yet soft, anchored yet minimal. It’s a vision of design that holds form without force.
Mapping the Drift
When style moves, where does it go?
This chart doesn’t follow fabric or print. It follows form. Each dot represents the average silhouette of a season from Issey Miyake’s archive, positioned in a shared visual space. The closer two points are, the more similar their structural composition; the further apart, the more a new silhouette language emerged.
Certain moments stay near each other—SS1995 and SS1998 speak in similar proportions, loose volumes, and soft geometry. Others pull away. SS2012 stands apart, marking a distinct structural shift, while SS2025 signals another break: cleaner lines, architectural precision, and a quieter force.
What the algorithm captures isn’t aesthetic judgment or outside opinion. It analyzes pixels to surface spatial relationships between collections. No narrative. Just position. And sometimes? That’s all you need.
About the author
Jack Iorio is a data-driven fashion writer based in New York. Through Ctrl + Couture, he blends data science and editorial analysis to uncover hidden patterns in fashion.