Inspiration
In Nigeria, one of the biggest frustrations in the fashion world is the “What I ordered vs What I got” experience. Customers send measurements or outfit ideas to tailors, especially from long distances, and end up receiving clothes that don’t fit or look nothing like what they imagined. We wanted to fix that by using AI to bring accuracy, visualization, and confidence into the tailoring process. That’s how PatternFit AI was born - a virtual tailor that helps users visualize their custom outfits and ensures perfect fitting before a single fabric is cut.
What it does
PatternFit AI is an AI-powered web app that simulates a full tailoring experience — virtually. Users can:
Choose their gender, style (Agbada, Ankara gown, etc.), sleeve length, and fabric.
Upload or take a full-body photo.
Upload custom fabric images for personalization.
Watch the AI estimate their measurements, analyze their choices, and generate a hyper-realistic virtual try-on of the outfit — showing exactly how it would look on them.
It also provides an AI Tailor’s Assessment — a detailed review that includes a fit analysis, fabric suitability, and style tips. Every creation is saved in a personal “Virtual Wardrobe” that users can revisit anytime.
How we built it
Frontend: React + TypeScript for a fast, modular interface.
Styling: Tailwind CSS for clean, responsive, and modern design.
AI Models:
gemini-2.5-flash-image for generating the virtual try-on images.
gemini-2.5-flash for structured text analysis and the AI tailor’s feedback.
State Management: React Hooks (useState, useEffect, useCallback).
Storage: localStorage for saving user fabrics and wardrobes.
Web APIs: Camera (getUserMedia), FileReader, Canvas (for compression), and Web Share for direct sharing.
Everything runs smoothly as a single-page application — no backend required.
Challenges we ran into
Getting realistic outfit blending while keeping the user’s face, body shape, and background unchanged was tough.
Prompt-tuning the AI for consistent try-on quality took several iterations.
Managing localStorage limits while storing multiple image generations required custom compression logic.
Ensuring good UX across different browsers and camera permissions.
Accomplishments that we're proud of
Achieved photo-realistic outfit generation directly in the browser.
Designed a smooth, three-step user flow that feels like a real tailor experience.
Built a completely offline-compatible wardrobe system using browser storage.
Created a solution that’s culturally relevant, empowering tailors and customers alike.
What i learned
How to integrate multimodal AI models into real-world web apps.
The importance of prompt engineering and schema design for consistent AI output.
That small UX details (like camera access and progress feedback) make a huge difference in user trust.
Collaboration between design, AI, and traditional fashion can open new possibilities for creative industries.
What’s next for PatternFit AI: The Automated Tailor
Add 3D body visualization for even more accurate virtual fitting.
Build a Tailor Dashboard, where tailors can receive client specs directly with auto-generated measurements.
Launch a mobile version (PWA) for broader access.
Partner with local designers and fashion schools to create a digital fitting ecosystem.
Built With
- built-with-react
- by
- gemini
- powered
Log in or sign up for Devpost to join the conversation.