12 Categories of Image Generation in awesome-gpt-image-2: Complete Industrial Guide

The awesome-gpt-image-2 repository organizes over 500 industrial-grade prompt cases into 12 core I2 (Industrial Image-generation) categories spanning UI design, data visualization, photography, and classical Chinese themes, with definitions stored in data/style-library.json and templates documented in docs/templates.md.

The freestylefly/awesome-gpt-image-2 project provides a structured taxonomy for production-ready image generation. These 12 categories of image generation map specialized prompt templates to distinct visual domains, enabling developers to programmatically retrieve reusable patterns for GPT-4o and DALL-E 3 workflows through a unified JSON API.

The 12 Core I2 Categories

The repository defines 12 primary industrial categories that classify every prompt case in the library. Each category targets a specific visual output type and includes dedicated template sections in docs/templates.md.

• 🧩 UI & Interfaces: Component hierarchies, page layouts, and screenshot textures. Covers everything from button states to full dashboard mockups.

• 📊 Charts & Infographics: Data visualization modules, directional arrows, and structural diagrams optimized for readability and information density.

• 📰 Posters & Typography: Headline systems, layout grids, and high-impact visual compositions for print and digital campaigns.

• 🛍️ Products & E‑commerce: Product detail pages, packaging mockups, and selling-point visualizations for retail workflows.

• 🏷️ Brand & Logos: Identity systems, logo variations, and brand touchpoint templates for consistent corporate visual language.

• 🏛️ Architecture & Spaces: Perspective rendering, material definitions, and indoor/outdoor lighting scenarios for built environments.

• 📷 Photography & Realism: Lens emulation, lighting setups, and texture workflows that replicate camera-based photographic output.

• 🎨 Illustration & Art: Brushwork definitions, material simulations, and stylistic frameworks for non-photorealistic rendering.

• 🧍 Characters & People: Character design sheets, pose consistency frameworks, and anthropomorphic figure generation.

• 🎬 Scenes & Storytelling: Storyboard sequences, world-building environments, and narrative pacing controls for sequential visual content.

• 🏮 History & Classical Chinese Themes: Dynasty-specific aesthetics, traditional clothing, and scroll-style narrative layouts for cultural heritage projects.

• 📚 Documents & Publishing: Page systems, table of contents generators, and typographic layout rules for structured documents.

The repository also includes a 🧪 Other Use Cases category for experimental workflows and mixed tasks that do not fit the primary taxonomy.

Source Files Defining the Category System

The category architecture is implemented across four critical files in the codebase:

• README.md: Provides the human-readable Category Overview with visual covers and case counts (e.g., "🧩 UI & Interfaces – 73 cases").

• docs/templates.md: Contains the full Prompt Template Categories with anchor links (e.g., #tpl-ui, #tpl-architecture) that map directly to the 12 categories.

• data/style-library.json: The machine-readable source of truth where every entry includes a category field linking it to one of the 12 taxonomies.

• agents/skills/gpt-image-2-style-library/SKILL.md: Documents the Agent skill API that exposes these categories programmatically.

Programmatic Access to Image Categories

You can retrieve the 12 categories of image generation programmatically via the bundled style library or the published NPM skill.

Node.js Implementation

Read the canonical category list directly from data/style-library.json:

import fs from 'fs';
import path from 'path';

// Resolve the style library that powers both the website and Agent skill
const libPath = path.resolve('data/style-library.json');
const lib = JSON.parse(fs.readFileSync(libPath, 'utf-8'));

// Extract unique I2 categories from the dataset
const categories = [...new Set(lib.map(entry => entry.category))];
console.log('I2 categories:', categories);

CLI Interface

If you have installed the gpt-image-2-style-library skill, list categories via the command line:

# Query the skill's category API

npx gpt-image-2-style-library list-categories

Both methods reference the same data/style-library.json source file, ensuring consistency between the web UI, documentation, and automated agent workflows.

Summary

• 12 primary I2 categories structure the awesome-gpt-image-2 repository's 500+ prompt cases into logical visual domains.
• Category definitions reside in data/style-library.json and render in docs/templates.md with specific anchor links.
• Programmatic access is available via Node.js file system reads or the gpt-image-2-style-library CLI tool.
• Industrial coverage spans UI, infographics, architecture, photography, illustration, characters, storytelling, and specialized cultural themes.

Frequently Asked Questions

What does I2 stand for in the awesome-gpt-image-2 repository?

I2 stands for Industrial Image-generation, a taxonomy designed for production-grade visual workflows rather than experimental art generation. The 12 I2 categories prioritize reusability, consistency, and functional output for commercial design tasks.

How are the 12 categories stored and maintained in the codebase?

Each prompt case in data/style-library.json includes a category field that maps it to one of the 12 taxonomies. This JSON file serves as the single source of truth for both the documentation site and the Agent skill, ensuring that category counts and definitions remain synchronized across all interfaces.

Can I extend the category system with custom prompt templates?

Yes. To add new templates, append entries to data/style-library.json using one of the existing 12 category strings, or define a new value if extending the taxonomy. The documentation in docs/templates.md provides the anchor link structure (e.g., #tpl-product) required for template organization.

How do I filter prompts by category when using the Agent skill?

The gpt-image-2-style-library skill exposes a filtering interface that accepts a category parameter. According to agents/skills/gpt-image-2-style-library/SKILL.md, you can pass any of the 12 category names to retrieve only prompts matching that industrial domain, enabling targeted generation workflows.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}