Fashion Twin: A Virtual Try On

Designing Fashion Twin, a calm virtual try-on experience that turns body and colour data into personalised Stone Island outfit suggestions, developed at the School of Innovation.

Role

UI/UX Designer

Industry

Clothing

Duration

2 Months

Stage 1. Framing the Problem

Online fashion still runs on guessing, leading to low confidence and a lot of waste.

High return rates: Around 25–30% of apparel orders are returned, and a large share is linked to size, fit or cut issues.
Opaque recommendations: Most fashion recommenders behave like black boxes; they don’t provide explainable recommendations about why an item should suit a specific body or colouring, so shoppers still rely on charts and model photos to imagine the fit.

Stage 2. Design Strategy

To replace guessing with guided decisions, I structured Fashion Twin around three layers in a minimal, luxury UI:

The Body – 3D Twin: A virtual twin (from manual input, photo or live capture) lets users see outfits on their own body instead of generic models.
The Style – Colour & Cut Logic: Light seasonal colour and body-shape cues steer them toward pieces that are more likely to fit and flatter.
The Choice – Explainable Recommendations: Each suggestion includes a short “why this works for you”, turning a black-box recommender into something transparent and easier to trust.

Stage 1. Framing the Problem

Online fashion still runs on guessing, leading to low confidence and a lot of waste.

High return rates: Around 25–30% of apparel orders are returned, and a large share is linked to size, fit or cut issues.
Opaque recommendations: Most fashion recommenders behave like black boxes; they don’t provide explainable recommendations about why an item should suit a specific body or colouring, so shoppers still rely on charts and model photos to imagine the fit.

Stage 2. Design Strategy

To replace guessing with guided decisions, I structured Fashion Twin around three layers in a minimal, luxury UI:

The Body – 3D Twin: A virtual twin (from manual input, photo or live capture) lets users see outfits on their own body instead of generic models.
The Style – Colour & Cut Logic: Light seasonal colour and body-shape cues steer them toward pieces that are more likely to fit and flatter.
The Choice – Explainable Recommendations: Each suggestion includes a short “why this works for you”, turning a black-box recommender into something transparent and easier to trust.

Creating Your Twin

This flow turns sensitive body data into a recognisable twin through three short paths (Manual, Photo, Live), each with a clear progress bar.

Low pressure: Users choose how they share their body, reducing friction and anxiety.
One step at a time: Each screen asks for a single focused input instead of a long form.
Instant feedback: The twin preview updates with each change, making the impact of inputs easy to see.

Explainable recommendations from your Twin

Once the twin is created, Fashion Twin shifts from input to feedback: it analyses body shape and colouring, gives a short summary of what tends to work, and turns that into a personal collection. This makes recommendations more transparent and explainable instead of a black box, and helps users build a clear mental model of why items are being suggested.

Explainable recommendations from your Twin

Trying on products with your Twin

On the product page, Fashion Twin replaces generic model shots with the user’s 3D twin and updates the avatar with every size and colour change, so fit decisions are made visually on their own body instead of guessing from a size chart.

Direct manipulation: Size and colour controls update the twin, not just a value in a dropdown.
Lower cognitive load: Users read fit from posture and proportions, rather than abstract size tables.
Brand consistency: The layout keeps Stone Island’s clean product page structure, only swapping the model image for the twin.

Outcomes & Reflections

Fashion Twin started as a start-up concept at the School of Innovation and grew into an award-winning project, which pushed me to think beyond screens and treat the work like a real product, not just a UI exercise.

Chunking information works: Breaking body data, style insight and try-on into small, separate steps felt lighter and more usable than a single “all-in-one” flow.
From UX project to start-up thinking: Working on Fashion Twin as a start-up idea taught me to balance desirability, feasibility and viability—pitching the concept, refining it with feedback and keeping the scope realistic.
Translating UX across domains: I learned to apply my UX process in a new area and keep decisions grounded in clarity and behaviour, not just visuals.
User-first, not feature-first: I became stricter about starting from the user’s mental state (vulnerability, uncertainty, decision fatigue) and letting that drive the flow, rather than the technology.

Other projects

Robot Fleet Operations Dashboard

Designing a clear, real-time dashboard to monitor and optimize a fleet of cleaning robots for Telmekom — winner of the NOI Hackathon SFSCON 2025.

ZenFlow: A Productivity & Task Manager App

From qualitative research (observations, interviews, focus group) to a zen-like planner UI that supports focus, progress awareness and gentle guidance.

Foodiebox: A Meal Box App

Foodiebox is designed for busy people who still want to cook at home, reducing the effort of choosing meals, buying ingredients and following recipes.