AI Room Renovator

Turn distressed rooms into listing-ready interiors using AI-powered virtual staging.

Overview

AI Room Renovator is an MVP web application designed for US realtors who need to virtually stage empty or outdated rooms for property listings. Upload a room photo, choose a design style, and receive a photorealistic renovation — all while preserving the original wall geometry via ControlNet.

Key Features

• Before / After Slider — interactive comparison using img-comparison-slider
• Multiple Interior Styles — Modern, Industrial, Minimalist (admin-managed)
• AI Generation — Replicate + ControlNet (geometry-preserving)
• Admin Panel — full CMS via AdminJS (users, styles, content, analytics)
• No Signup Required — instant upload-and-generate workflow
• Responsive — mobile-first layout with Tailwind CSS

Architecture

prop2ai/
├── server.js              # Express server, API routes, boot sequence
├── models/index.js        # Sequelize models (SQLite) — User, Style, etc.
├── admin/index.js         # AdminJS configuration & resources
├── services/v0Client.js   # v0 API integration for design assistance
├── public/
│   ├── index.html         # Single-page frontend (Tailwind CDN)
│   └── script.js          # Client-side logic & API interactions
├── .env                   # Environment variables (not committed)
└── data.db                # SQLite database (auto-created, not committed)

Tech Stack

Layer	Technology
Backend	Node.js 20+, Express 4, ESM
Database	SQLite via Sequelize 6
Admin	AdminJS 7 + @adminjs/sequelize
AI Engine	Replicate API (ControlNet models)
Frontend	Vanilla JS, Tailwind CSS (CDN)
Slider	img-comparison-slider 8
Design	v0 Model API (design assistance)

Quick Start

Prerequisites

• Node.js ≥ 20
• A Replicate API token

Installation

git clone https://github.com/offflinerpsy/prop2ai.git
cd prop2ai
npm install

Configuration

Create a .env file in the project root:

REPLICATE_API_TOKEN=r8_your_token_here
V0_API_KEY=v0_your_key_here          # optional — for design-assist endpoint
PORT=3000                             # optional — defaults to 3000
SESSION_SECRET=change-me
ADMIN_COOKIE_SECRET=change-me

Running

node server.js

• App: http://localhost:3000
• Admin Panel: http://localhost:3000/admin Default login: admin@prop2ai.com / admin123

API Endpoints

Method	Path	Description
GET	/health	Health check
GET	/api/styles	List active interior styles
GET	/api/blocks	List visible content blocks
GET	/api/menu	Navigation menu links
GET	/api/settings/public	Non-secret site settings
GET	/api/stats	Dashboard statistics
POST	/api/restage	Generate renovated room image
POST	/api/design-assist	Get v0 design suggestions

POST /api/restage

curl -X POST http://localhost:3000/api/restage \
  -F "image=@room.jpg" \
  -F "user_style=Modern" \
  -F "structure=hough"

Response: { "outputUrl": "https://replicate.delivery/..." }

Admin Panel

AdminJS provides a full CRUD interface at /admin:

Section	Description
Users & Access	Admin & client accounts, role management
AI Styles	Prompts, negative prompts, preview images, ordering
Content Blocks	Hero, trust cards, CTA blocks with icons & config
Navigation	Header menu links with visibility control
Settings	API keys, branding, site configuration
Generations	Full AI generation log (status, timing, errors)

Adding a New Style

1. Navigate to /admin → AI Styles → Create new
2. Fill in: name, slug, prompt, negative prompt, preview image URL
3. Set isActive: true and order
4. The style card appears on the frontend automatically

Database

SQLite stored at data.db (auto-created on first run). No external DB required.

Models: User, Style, SiteBlock, Setting, Generation, MenuLink, ReplicateModel.

Security

• Never commit .env or data.db
• Change the default admin password after first login
• API tokens can be managed from Admin → Settings → API Keys
• Session secrets should be changed in production

Contributing

1. Fork the repository
2. Create a feature branch: git checkout -b feature/my-feature
3. Commit with conventional messages: feat: add style preview
4. Push and open a Pull Request

License

MIT