Inspiration
Interior redesigning is often expensive, time-consuming, and difficult to visualize before committing money. We wanted to build LuminaSpace AI to bridge the gap between imagination and reality—allowing anyone to upload a room photo and instantly generate personalized, high-quality redesigns paired with real-time shopping recommendations and budget management.
What it does
LuminaSpace AI transforms interior design into a seamless, interactive workflow:
- AI Room Generation: Renders photorealistic room redesigned variants from user-uploaded images using advanced diffusion models.
- Smart Budget Management: Integrates a real-time budget tracking tool ($ \text{Cart Total} \le \text{Budget} $) that lets users monitor expenses as they shop for catalog furniture.
- Dynamic API Failover: Utilizes a custom Hugging Face multi-key rotation mechanism to guarantee uninterrupted AI image generation during rate limits or high usage.
How we built it
- Frontend: Built with React, Vite, and Tailwind CSS for a fast, modern responsive dark-mode UI.
- Backend & Auth: Powered by Supabase for database state management and secure Google OAuth integration.
- AI Pipeline: Integrated Google Gemini 3.7 alongside Hugging Face inference models for image processing and key rotation.
- Deployment: Continuous Integration/Deployment configured and hosted on Vercel.
Challenges we faced
- API Rate Limits: Handling high traffic and API request quotas was tricky. We solved this by engineering an automated multi-key rotation fallback system for Hugging Face endpoints.
- Real-time State Sync: Ensuring immediate layout re-renders between user auth states, room selections, and cart budget updates required tight state management.
What we learned
- Designing robust failover architectures for third-party AI APIs.
- Streamlining complex UI controls (budget tracking, model badges, catalog items) into an intuitive, seamless user interface.
What's next for LuminaSpace AI
- Integrating 3D interactive room previews.
- Direct e-commerce checkout integration for catalog items.
Built With
- css3
- gemini-api
- git
- html5
- huggingface
- javascript
- react
- supabase
- tailwindcss
- vercel
- vite
Log in or sign up for Devpost to join the conversation.