Назад към всички

pollinations-image

// A clean, modular, and user-friendly image generation tool powered by [Pollinations.ai](https://pollinations.ai).

$ git log --oneline --stat
stars:1,933
forks:367
updated:March 4, 2026
SKILL.mdreadonly

🎨 Pollinations Image Generation Skill

A clean, modular, and user-friendly image generation tool powered by Pollinations.ai.

✨ Features

  • 🎁 5,000 free images/month with pollen grants
  • 🔄 Easy model switching - change anytime
  • 💾 Persistent defaults - remember your preferences
  • 📊 Clear model comparison - quality, speed, cost
  • 🔧 Modular design - easy to extend

🚀 Quick Start

First Time Setup

# 1. Clone or copy the skill
cd pollinations-image/

# 2. (Optional) Configure your API key
cp .env.example .env
# Edit .env and add your key from https://enter.pollinations.ai

# 3. (Optional) Set your default model  
cp .user.conf.example .user.conf
# Edit .user.conf to change DEFAULT_IMAGE_MODEL

# 4. List all available models
./generate.sh models

# 5. Generate your first image
./generate.sh "a cute purple cat"

Already Configured?

# List all available models
./generate.sh models

# Set your default model (optional)
./generate.sh set-model flux

# Generate an image
./generate.sh generate --prompt "a cute purple cat"

# Or use the shortcut
./generate.sh "a cute purple cat"

📋 Available Models

ModelTypeSpeedQualityCost
flux🎁 Free⚡⚡⭐⭐⭐⭐⭐~0.0002 pollen/img
zimage🎁 Free⭐⭐⭐⭐~0.0002 pollen/img
klein💰 Paid⚡⚡⭐⭐⭐⭐⭐~0.008 pollen/img
klein-large💰 Paid⚡⚡⚡⭐⭐⭐⭐⭐⭐~0.012 pollen/img
gptimage💰 Paid⚡⚡⭐⭐⭐⭐⭐2.0-8.0 pollen/M

📖 Commands

generate (or g)

Generate an image with your prompt.

# Full command
./generate.sh generate --prompt "a sunset over mountains" --model flux --width 1024 --height 1024

# Short options
./generate.sh g -p "a sunset" -m zimage -w 512 -h 512

# Even shorter (prompt only)
./generate.sh "a cute cat"

Options:

  • --prompt, -p - Image description (required)
  • --model, -m - Model to use (overrides default)
  • --width, -w - Image width (default: 1024)
  • --height, -h - Image height (default: 1024)
  • --seed, -s - Random seed for reproducibility
  • --filename, -f - Custom output filename
  • --nologo - Remove Pollinations watermark
  • --enhance - Let AI improve your prompt

models (or m)

Display all available models with ratings.

./generate.sh models

model MODEL_NAME

Show detailed information about a specific model.

./generate.sh model klein-large

set-model MODEL_NAME

Set your default model for all future generations.

./generate.sh set-model zimage

Note: This saves to .user.conf and persists across sessions.

config

Display your current configuration.

./generate.sh config

help

Show help message.

./generate.sh help

⚙️ Configuration

User preferences are stored in .user.conf:

# Example .user.conf
DEFAULT_IMAGE_MODEL=flux

API keys are stored in .env (not tracked in git):

# Example .env
POLLINATIONS_API_KEY=your_key_here

📁 Structure

pollinations-image/
├── generate.sh        # Main entry point
├── lib/
│   └── models.sh      # Model registry and metadata
├── .env               # API keys (private)
├── .user.conf         # User preferences
└── SKILL.md           # This documentation

🎯 Recommended Workflows

Draft → Refine → Final

# 1. Quick draft with fast model
./generate.sh g -p "concept sketch of a dragon" -m zimage -w 512

# 2. Refine with better model
./generate.sh g -p "detailed dragon in a castle" -m flux -w 1024

# 3. Final high-quality render
./generate.sh g -p "masterpiece, highly detailed dragon..." -m klein-large

Daily Driver Setup

# Set fast model for daily use
./generate.sh set-model zimage

# All future generations use zimage by default
./generate.sh "quick concept"

# Override for special occasion
./generate.sh "important artwork" --model klein-large

🔑 Getting an API Key

  1. Visit enter.pollinations.ai
  2. Create an account
  3. Request a pollen grant (free tier: 5K images/month)
  4. Generate an API key
  5. Save it to .env: POLLINATIONS_API_KEY=your_key

🤝 Contributing & Sharing

This skill is designed to be shared!

Files to Include

When sharing the skill, include:

  • generate.sh — Main script
  • lib/models.sh — Model registry
  • .env.example — Example API key file
  • .user.conf.example — Example user config
  • SKILL.md — Documentation

Files to Exclude (Private)

Do NOT share these (they contain personal data):

  • .env — Your private API key
  • .user.conf — Your personal preferences
  • .first-run-complete — Setup marker

Adding a New Model

  1. Edit lib/models.sh
  2. Add your model to MODELS_LIST following the format:
    model_id|Display Name|type|cost|speed|quality|Description
    
  3. Test with ./generate.sh model your_model_id

First-Time User Experience

When someone first runs the skill without config files, they'll see a welcome message guiding them through setup. The skill works out-of-the-box with sensible defaults:

  • Default model: flux (free, high quality)
  • API key: Optional (works with public endpoint)

📝 License

MIT - Feel free to use, modify, and share!


Made with 🫐 for the OpenClaw community.