AYNA – AI Virtual Try-On Atelier
AI-powered South Asian bridal virtual try-on and in-store smart-mirror platform with interactive 3D visualization.
3D Graphics & Computer Vision Developer
Team of 3 (Build With AI Hack Days)
Build With AI Hack Days @ SEU
Generative AI & 3D Web
March 2026
Multimodal AI + 3D
The Problem & Real-World Friction
Bridal fashion sizing and physical try-ons in South Asian apparel require significant fitting time, with physical inventory limitations in physical stores.
Architectural & Technical Approach
Developed an AI smart-mirror kiosk combining webcam input, multimodal AI garment visualization, QR-code result transfer to mobile, automated privacy session reset, AI product onboarding, and interactive 3D model visualization.
Key Technical Highlights
AI Smart-Mirror & 3D Garment Atelier
Multimodal AI Garment Transfer
Google Gemini multimodal vision pipeline mapping traditional South Asian bridal garments onto live user camera feeds in real-time.
Interactive 3D Atelier Rendering
Embedded Three.js and React Three Fiber 3D viewport allowing customers to inspect garment textiles and embroidery from any 360° angle.
Zero-Trace Privacy Auto-Reset
Smart mirror kiosk automatically flushes local camera buffer and session cache on user departure to guarantee complete customer privacy.
QR Bridge to Mobile Handset
One-scan QR code bridge allowing shoppers to instantly save high-res try-on renders to their personal mobile device without signing in.
Project Results & Impact
Built for Build With AI Hack Days at Southeast University, showcasing fast AI garment transfer and real-time 3D styling.
Reproducible Benchmarks & Test Setup
Test setup, evaluation procedure, and validated outcomes reported for this architecture:
March 2026
Local WebGL runtime profiling on Apple M-series and Intel Iris Xe GPUs.
Outcome: Stable 60 FPS in 3D viewport. 2.3s avg generative transfer turnaround.
Separation of Implemented, Prototyped & Planned Features
- •Three.js 3D interactive mannequin with 360° rotation and textile inspection
- •Webcam body silhouette segmentation via MediaPipe
- •Zero-trace local buffer flush after 60s idle timeout
- •Multimodal generative cloth drape transfer via Gemini 1.5 Pro edge proxy
- •QR code handoff to mobile browser
- •Multi-person retail kiosk queue management
Role & Contributions
3D Graphics & Computer Vision Developer
Team of 3 (Build With AI Hack Days)
- Engineered Three.js 3D viewport with PBR shaders for traditional South Asian bridal fabric textures.
- Implemented MediaPipe 33-point body landmark alignment for live camera segmentation.
- Integrated Gemini 1.5 Pro multimodal synthesis with zero-trace local memory reset for customer privacy.
Technologies Used
7 toolsCivic AI
AI-powered civic issue reporting and government management platform that won Champion at the AI & API Hackathon 2026.
View next case studyInterested in this architecture?
Have a question about the system design, tech stack, or potential collaboration?
Let's discuss