Paving Design System — AI-Powered Stone Pattern Generator

Automated paving design pipeline that generates architectural stone patterns, photorealistic renders, bill of materials, and installation maps. Built with Streamlit, Stable Diffusion + ControlNet, and algorithmic pattern engines.

What It Does

Given a site photo and design preferences, the system:

1. Renders photorealistic paving visualization using SD + ControlNet (maintains architectural perspective)
2. Generates algorithmic patterns (disintegration, gradient, halftone, pixel art)
3. Calculates Bill of Materials (paver count + m² per color)
4. Exports IKEA-style installation grid maps
5. Produces DXF files for CAD integration

Example Projects

Terrace Entrance	Garden Pathway	Plaza Center

Garden Corner	Entry Foyer

Architecture

Site Photo → Depth Map (MiDaS) → SD + ControlNet Render
                                         ↓
User Config → Pattern Engine → Grid Map → BoM Calculator
         (area, paver size,       ↓            ↓
          pattern style)    Installation   Material
                              Map          Report
                               ↓
                          DXF / HTML Export

Pattern Engine

• Disintegration — radial probability field, dense center dissolving outward
• Gradient — smooth color transitions across the surface
• Halftone — circular dot patterns scaled by intensity
• Pixel Art — grid-based color mapping

Render Pipeline (SD + ControlNet)

• Depth map extraction via MiDaS preserves architectural perspective
• ControlNet conditioning forces pattern to wrap around surface geometry
• Not a flat overlay — pattern respects slope, foreshortening, and grout lines

Usage

pip install -r requirements.txt

# Run web app
streamlit run app.py

# Generate example projects
python generate_examples.py

# Full pipeline (programmatic)
python paving_pipeline.py

Tech Stack

Layer	Technology
Web UI	Streamlit
Rendering	Stable Diffusion + ControlNet (RunPod API)
Pattern Engine	NumPy, Pillow
CAD Export	DXF (ezdxf)
Storage	Local filesystem / Cloudflare R2
Database	SQLite
Deployment	Docker, nginx

Project Structure

├── app.py                 # Streamlit web application
├── core/
│   ├── patterns.py        # Pattern generation algorithms
│   ├── pattern_renderer.py # Pattern → image rendering
│   ├── grid_renderer.py   # Installation grid map
│   ├── bom.py             # Bill of Materials calculator
│   ├── colors.py          # Color management (97 paver colors)
│   ├── dxf_export.py      # CAD DXF export
│   ├── html_export.py     # HTML report export
│   ├── render_client.py   # SD + ControlNet API client
│   ├── storage.py         # File storage (local / R2)
│   ├── db.py              # Project database
│   ├── i18n.py            # Internationalization (TR/EN)
│   └── prompt_builder.py  # SD prompt construction
├── deploy/                # Production deployment scripts
├── tests/                 # Unit tests
└── output/examples/       # 5 complete project examples