The Algorithmic Loom — IAT 460 Final Project

A generative art system that transforms written text into emotion-driven cross-stitch embroidery patterns. Input text is analyzed for its emotional content using Google Gemini, which extracts a primary emotion, intensity score, poetic subtitle, and a cohesive 7-color thread palette. These parameters seed a Perlin noise field that produces a 50×50 woven grid pattern, rendered both as an interactive canvas preview and a downloadable PDF blueprint.

Setup and Installation

1. Clone the repository and navigate into the project directory.
2. Create and activate a virtual environment (recommended):

   python -m venv venv
   source venv/bin/activate
   

3. Install dependencies:

   pip install -r requirements.txt
   

4. Create a .env file in the project root with your Gemini API key:

   GEMINI_API_KEY=your_key_here
   

5. Run the application:

   python app.py
   

6. Open http://127.0.0.1:5000/ in your browser.

How It Works

1. The user enters any piece of text — a poem, journal entry, memory, or stream of thought.
2. The text is sent to Google Gemini (gemini-2.5-flash), which returns structured emotional metadata: a primary emotion, intensity score (0.0–1.0), a 7-color palette sorted dark to light with poetic thread names, and an evocative subtitle.
3. A 50×50 Perlin noise grid is generated, deterministically seeded by the emotion string and intensity. Higher intensity produces more octaves (texture complexity) and wider palette distribution.
4. The grid is animated cell-by-cell onto an HTML canvas with contrast-aware center dots mimicking cross-stitch markings.
5. The user can export the result as a PDF blueprint containing the title, subtitle, emotion metadata, the full color grid, and a thread legend.

Project Structure

File	Description
app.py	Flask web server with routes for analysis and PDF rendering
llm.py	Gemini API integration with retry logic and structured JSON validation
loom.py	Perlin noise grid generation and ReportLab PDF blueprint rendering
static/templates/index.html	Single-page frontend with canvas rendering and animated UI
requirements.txt	Pinned Python dependencies
outputs/	Sample PDF blueprint outputs

Dependencies

• Flask — Web application framework
• python-dotenv — Environment variable management
• google-genai — Google Gemini API client
• perlin-noise — Perlin noise generation for the grid field
• reportlab — PDF document generation