AI-Based Interior Designer transforms room photos into stunning redesigns using a multi-stage AI pipeline. Upload an image, select style and room type, and receive three high-quality variants with detailed analysis.
- AI-Powered Pipeline: LLaVA analyzes room → Enhanced prompts → FLUX.1 generates 3 design variants per upload
- Interactive UI: Drag-and-drop upload, theme/room selectors, animated results reveal
- Full Analysis: Image descriptions, optimized prompts, processing times displayed
- Export Ready: Individual image downloads, copy-to-clipboard prompts
- Responsive Design: Glassmorphism UI with Framer Motion animations
| Component | Technology |
|---|---|
| Frontend | Next.js 15, TypeScript, Tailwind CSS, Framer Motion |
| UI Components | shadcn/ui, Heroicons, react-dropzone |
| AI Pipeline | LLaVA (Ollama), FLUX.1 (Hugging Face), Custom prompt enhancement |
| Animations | Staggered reveals, scroll-triggered effects |
-
Clone repository and install dependencies npm install
-
Copy environment template cp .env.example .env
-
Configure API keys (see Environment Variables)
-
Run Ollama with LLaVA model First Install Ollama, then: ollama pull llava ollama pull mistral ollama serve
-
Start development server npm run dev
Create .env.local using .env.example:
Hugging Face API token for FLUX.1 inference HF_TOKEN=your_huggingface_token_here
Ollama API endpoint (default localhost) OLLAMA_API_URL=http://localhost:11434
.env.example content: HF_TOKEN= OLLAMA_API_URL=http://localhost:11434
- Upload: Drag PNG/JPEG (<5MB) or click to select
- Configure: Select room type and design theme
- Optional: Add custom instructions (max 100 chars)
- Generate: Click submit → Watch 3 variants appear sequentially
- Export: Download images or copy enhanced prompts
POST /api/generate receives FormData with image, style, room, userPrompt.
Response Structure: { success: boolean, model: string, // "FLUX.1-schnell" pipeline: string, // "llava→enhance→flux" imageDescription: string,// LLaVA room analysis enhancedPrompts: string[], // 3 optimized variants images: string[], // 3 base64 PNGs (no data: prefix) processingTime: string, // "12.4s" originalImageBase64: string }
Themes: Modern, Vintage, Minimalist, Professional
Rooms: Living Room, Dining Room, Bedroom, Bathroom, Office
| Issue | Solution |
|---|---|
| "Network error" | Verify Ollama running at localhost:11434; check HF_TOKEN |
| No image generation | Ensure FLUX model accessible via HF_TOKEN |
| Upload rejected | PNG/JPEG only, <5MB limit |
| Slow generation | FLUX.1-schnell is fastest; upgrade hardware/GPU for dev/prod |
layout.tsx: EnhancedSidebar wrapper- Home page: Landing with features showcase
- Dashboard (
page.tsx): Core upload/generate UI - Custom hooks/components: UploadZone, GlassButton, SelectMenu, ActionPanel
Strict Proprietorship License - Automatech
Developed by Amaan MK (https://github.com/Amaan9136)
Owned by Automatech - All Rights Reserved 2025-2026 Copyright Act.