Prompt Builder - Advanced AI Prompt Generation

Professional-grade prompt generation for AI image creation with intelligent batch processing, anime styles, and advanced randomization

Overview

Prompt Builder is the most advanced AI prompt generation system available, designed for creators who demand professional results. Whether you're using the standalone web application or the ComfyUI integration, you get the same powerful features that transform simple descriptions into detailed, high-quality prompts.

---

Choose Your Version

Standalone Web Application (Main Branch)

Perfect for direct use, privacy-focused local processing, and development

• Beautiful Web Interface - Modern, responsive React-based UI
• Complete Privacy - All processing happens locally on your machine
• Local LLM Integration - Works with Ollama, LM Studio, Text Generation WebUI
• Real-time Generation - Instant prompt creation and live preview
• Advanced Character Builder - Detailed customization with live preview
• 20+ Anime Styles - From Studio Ghibli to modern anime
• Intelligent Batch Processing - Generate 1-100 prompts with smart randomization
• History & Favorites - Save and manage your best prompts
• Developer Ready - Full TypeScript/React source code
• Built-in Analytics - Prompt analysis and optimization suggestions
• Multiple Deployment Options - Web app, Electron desktop, or Docker

Download Standalone App

ComfyUI Integration (ComfyUI-Node Branch)

Seamless integration with your ComfyUI workflows

• Native Integration - Works directly in ComfyUI
• Workflow Ready - Drag & drop nodes
• Batch Processing - Generate 1-100 prompts at once
• Anime Styles - 20+ specific anime art styles

Install ComfyUI Nodes

---

Core Features (Both Versions)

Intelligent AI Generation

• Local LLM Support - Ollama, LM Studio, Text Generation WebUI
• Online APIs - OpenAI, Claude, Gemini, Groq, DeepSeek, and more
• Smart Prompting - Context-aware prompt enhancement
• Model Optimization - Tailored for each AI model

Advanced Character Control

• Physical Attributes - Gender, age, body type, ethnicity
• Detailed Customization - Height, build, facial features
• NSFW Support - Three modes with granular control
• Preserved Traits - Custom features that never change

Professional Style System

• Realistic Styles - Professional, amateur, flash photography
• Anime Styles - 20+ specific styles (Ghibli, Naruto, Bleach, etc.)
• Dynamic Selection - Context-aware style options
• Quality Enhancement - Automatic model-specific tags

Intelligent Batch Processing

• Smart Randomization - Fixed vs. variable elements
• Batch Generation - 1-100 prompts with one click
• Character Consistency - Same person, different scenarios
• Selective Variation - Choose what randomizes

---

ComfyUI Installation

Method 1: ComfyUI Manager (Recommended)

1. Open ComfyUI
2. Click "Manager" → "Install Custom Nodes"
3. Paste: https://github.com/btitkin/promptbuilder.git
4. Select branch: comfyui-node
5. Click "Install" → Restart ComfyUI

Method 2: Manual Installation

# Navigate to ComfyUI custom nodes folder
cd ComfyUI/custom_nodes/

# Clone the ComfyUI branch
git clone -b comfyui-node https://github.com/btitkin/promptbuilder.git

# Install dependencies
cd promptbuilder
pip install -r requirements.txt

# Restart ComfyUI

ComfyUI Nodes Overview

Available Nodes

Main Generation Nodes

• Prompt Builder (Local LLM) - Works with local AI models (Ollama, LM Studio, etc.)
• Prompt Builder (Online LLM) - Supports major online APIs (OpenAI, Claude, Gemini, etc.)
• Quick Preset & Batch - Simplified interface with batch processing

Display & Utility Nodes

• Prompt Text Display - Shows prompts in ComfyUI interface
• Show Text - Display any text directly in nodes
• Prompt Selector - Choose and customize specific outputs
• Prompt Display & Stats - Formatted display with statistics

Complete Feature Set

• Model Support - SDXL, Pony, Flux, Illustrious, NoobAI, MidJourney, and more
• Character Controls - Gender, age, body type, ethnicity, detailed attributes
• NSFW Support - Three modes: Off, NSFW, Hardcore with granular controls
• Anime Styles - 20+ specific styles (Ghibli, Naruto, Bleach, One Piece, etc.)
• Preset System - Shot, Pose, Location, and Clothing presets
• Quality Tags - Automatic model-specific enhancement
• Batch Processing - Generate 1-100 prompts with intelligent randomization
• Smart Randomization - Fixed character + variable scenes
• Performance Caching - Faster generation with smart caching

---

Standalone Web Application

Key Features

Privacy-First Design

• 100% Local Processing - No data sent to external servers
• Offline Capable - Works without internet connection
• Your Data Stays Yours - Complete control over your prompts and settings

Advanced Local LLM Integration

• Ollama Support - Direct integration with Ollama models
• LM Studio Compatible - Works with LM Studio's OpenAI API
• Text Generation WebUI - Supports oobabooga's text-generation-webui
• Custom APIs - Any OpenAI-compatible local API
• Model Flexibility - Use any model size that fits your hardware

Professional Prompt Engineering

• Smart Character Builder - Detailed physical and personality traits
• Style System - Realistic photography + 20+ anime art styles
• Quality Enhancement - Automatic model-specific optimization
• Batch Generation - Create 1-100 variations with intelligent randomization
• Preset Library - Pre-built combinations for common scenarios

Developer Experience

• Modern Tech Stack - React 18, TypeScript, Vite
• Component Library - Reusable UI components
• API Abstraction - Easy to add new LLM providers
• Hot Reload - Instant development feedback
• Build Optimization - Production-ready builds

Quick Start Guide

Step 1: Clone and Install

# Clone the main branch
git clone https://github.com/btitkin/promptbuilder.git
cd promptbuilder

# Install dependencies
npm install

Step 2: Setup Local LLM (Choose One)

Option A: Ollama (Recommended)

# Install Ollama
curl -fsSL https://ollama.ai/install.sh | sh

# Download a model
ollama pull mistral
# or
ollama pull llama2

# Start Ollama (runs on http://localhost:11434)
ollama serve

Option B: LM Studio 1. Download and install LM Studio 2. Download a model (e.g., Mistral 7B, Llama 2 7B) 3. Start local server (runs on http://localhost:1234)

Option C: Text Generation WebUI

# Clone and setup text-generation-webui
git clone https://github.com/oobabooga/text-generation-webui
cd text-generation-webui
# Follow their installation guide
# Start with --api flag for API access

Step 3: Start the Application

# Development mode (with hot reload)
npm run dev

# Production build
npm run build
npm run preview

Step 4: Configure LLM Connection

1. Open the app in your browser
2. Go to Settings → API Configuration
3. Set your LLM endpoint:
4. Ollama: http://localhost:11434
5. LM Studio: http://localhost:1234
6. Text Gen WebUI: http://localhost:5000
7. Select your model name
8. Test connection and start generating!

Deployment Options

Desktop Application (Electron)

# Build desktop app
npm run electron:build

# Development with Electron
npm run electron:dev

Docker Deployment

# Build Docker image
docker build -t promptbuilder .

# Run container
docker run -p 3000:3000 promptbuilder

Static Hosting

# Build for static hosting
npm run build

# Deploy to Netlify, Vercel, GitHub Pages, etc.
# Upload the 'dist' folder

Configuration Options

Environment Variables:

# .env file
VITE_DEFAULT_API_URL=http://localhost:11434
VITE_DEFAULT_MODEL=mistral
VITE_ENABLE_ANALYTICS=false
VITE_THEME=dark

Custom API Providers:

// Add custom LLM provider in src/services/
export const customApiService = {
  generatePrompt: async (config, prompt) => {
    // Your custom API logic
  }
}

Example Use Cases

Character Consistency

Input: "A beautiful woman"
Settings: Fixed (female, 25s, curvy, blonde hair, green eyes)
Batch: 100 prompts with random locations/poses
Result: Same woman in 100 different scenarios

Anime Style Exploration

Style: Anime → Ghibli
Result: "Studio Ghibli style - soft, detailed, magical atmosphere"

Style: Anime → Naruto  
Result: "Naruto anime style - dynamic action poses, ninja themes"

Smart Batch Processing

Batch Count: 50
Mode: Fixed Character
Randomize: Locations ✓, Poses ✓, Clothing ✓
Preserved: "freckles, dimples, blue eyes"
Result: 50 unique scenes with consistent character

---

Technical Details

Supported Models

• SDXL - Stable Diffusion XL
• Pony - Pony Diffusion V6
• Flux - Black Forest Labs Flux
• Illustrious - Anime/Illustration focused
• NoobAI - Community anime model
• SD 1.5 - Classic Stable Diffusion
• MidJourney - MidJourney-optimized prompts

AI Providers

• Local: Ollama, LM Studio, Text Generation WebUI, Kobold AI
• Online: OpenAI, Claude, Google Gemini, Groq, DeepSeek, Together AI

Performance

• Single Prompt: ~1-3 seconds
• Batch (100): ~2-5 minutes (with caching)
• Memory Usage: <100MB
• Cache Hit Rate: 85%+ for similar requests

---

Documentation & Support

Resources

• Example Workflows - Ready-to-use ComfyUI workflows
• API Documentation - Complete API reference
• Configuration Guide - Setup and customization
• Style Guide - All available styles and presets

Community & Support

• Report Issues - Bug reports and feature requests
• Discussions - Community discussions
• Contact - Direct support

Contributing

• Pull Requests - Code contributions welcome
• Documentation - Contribution guidelines
• Style Guide - Development standards

---

License & Credits

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• ComfyUI Team - For the amazing ComfyUI framework
• Stability AI - For Stable Diffusion models
• Community Contributors - For feedback, testing, and improvements
• AI Model Creators - For the incredible AI models we support

Show Your Support

If you find Prompt Builder useful, please consider: - Starring this repository - Forking for your own projects - Sharing with the community - Reporting bugs and issues - Suggesting new features

---

Made with ❤️ for the AI Art Community

Transform your creative vision into perfect prompts

Back to Top