Image Gen

Image-Gen Skill

Production-grade AI image generation with DALL-E 3/2 support, cost tracking, and comprehensive style presets.

When to Use

✅ USE this skill when:

• Creating AI-generated images from text prompts
• Needing custom illustrations or concept art
• Generating multiple variations of images
• Creating visual content for presentations
• Designing graphics or mockups
• Needing AI art in specific styles
• Batch generating images for testing
• Needing programmatic image generation

❌ DON'T use this skill when:

• Editing existing images → Use image editing tools
• Creating vector graphics → Use design software
• Needing screenshots → Use screenshot tools
• Generating animations → Use video generation
• Needing stock photos → Use stock photo services

Prerequisites

# 1. Get OpenAI API Key
# Visit: https://platform.openai.com/api-keys
# Create and copy your API key

# 2. Set environment variable
export OPENAI_API_KEY="sk-your-api-key-here"

# 3. Verify
node --version

Commands

Basic Usage

# Generate image with default settings (DALL-E 3, 1024x1024)
{baseDir}/image-gen.js "A serene mountain landscape at sunset"

# Generate with specific model
{baseDir}/image-gen.js "Cyberpunk city" --model dall-e-3

# Generate multiple images
{baseDir}/image-gen.js "Abstract art" --n 4

# Output to specific directory
{baseDir}/image-gen.js "Ocean waves" --output-dir ./my-images

Model Selection

# DALL-E 3 - Best quality, $0.04/image
{baseDir}/image-gen.js "A photo of a cat" --model dall-e-3

# DALL-E 2 - Faster, $0.02/image
{baseDir}/image-gen.js "A photo of a cat" --model dall-e-2

Model Comparison:

Size Options

# DALL-E 3 sizes
{baseDir}/image-gen.js "Portrait of a person" --size 1024x1792    # Portrait
{baseDir}/image-gen.js "Landscape scene" --size 1792x1024          # Landscape
{baseDir}/image-gen.js "Square composition" --size 1024x1024       # Square

# DALL-E 2 sizes
{baseDir}/image-gen.js "Avatar" --model dall-e-2 --size 256x256     # Small avatar
{baseDir}/image-gen.js "Icon" --model dall-e-2 --size 512x512       # Icon size

Size Pricing (DALL-E 3):

Quality Settings

# Standard quality (default, faster)
{baseDir}/image-gen.js "A photo of a flower" --quality standard

# HD quality (slower, better detail)
{baseDir}/image-gen.js "A photo of a flower" --quality hd --size 1024x1792

Quality Comparison:

Visual Style

# Vivid style (more dramatic, default)
{baseDir}/image-gen.js "Fireworks display" --style vivid

# Natural style (more realistic)
{baseDir}/image-gen.js "Fireworks display" --style natural

Custom Style Presets

# Photographic
{baseDir}/image-gen.js "Lion portrait" --custom-style photographic
# Result: "A photorealistic image of Lion portrait, professional photography, 35mm, high detail, 8k"

# Digital Art
{baseDir}/image-gen.js "Fantasy castle" --custom-style digital_art
# Result: "A digital illustration of Fantasy castle, digital art, concept art, trending on artstation"

# Cinematic
{baseDir}/image-gen.js "Warrior in battle" --custom-style cinematic
# Result: "Cinematic scene of Warrior in battle, cinematic lighting, film grain, movie still, anamorphic"

# Watercolor
{baseDir}/image-gen.js "Sunset over mountains" --custom-style watercolor
# Result: "A watercolor painting of Sunset over mountains, wet-on-wet technique, soft colors, artistic"

Available Style Presets:

Batch Generation

# Generate 4 variations
{baseDir}/image-gen.js "Product photo of a wristwatch" --n 4

# Generate 10 variations (max)
{baseDir}/image-gen.js "Logo concept" --n 10

Batch Output:

/tmp/image-gen-2024-01-15/
├── gen_product-photo_1705312345678_1.png
├── gen_product-photo_1705312345678_2.png
├── gen_product-photo_1705312345678_3.png
├── gen_product-photo_1705312345678_4.png
└── index.html  (gallery view)

Features

Rate Limiting

Built-in protection against API limits:

• Default: 5 requests per minute
• 1 request per second minimum
• Automatic delays between requests

# First request
{baseDir}/image-gen.js "Image 1"

# Second request (automatically delayed)
{baseDir}/image-gen.js "Image 2"

Cost Tracking

Automatic cost calculation and reporting:

Total cost: $0.16
Images: 4
Session: 45s

DALL-E Pricing:

Gallery Generation

When generating multiple images, an HTML gallery is automatically created.

{baseDir}/image-gen.js "Abstract patterns" --n 4

Opens in browser with:

• Thumbnail grid
• Full-size preview on click
• Metadata for each image (prompt, size, cost)
• Dark theme styling

Error Handling

Content Policy Violations:

{baseDir}/image-gen.js "Inappropriate content"
# ❌ ERROR: Content policy violation: This prompt may violate OpenAI's content policy

Rate Limiting:

# After many requests:
{baseDir}/image-gen.js "Another image"
# ... waits and retries automatically

Payment Required:

{baseDir}/image-gen.js "Test image"
# ❌ ERROR: Pay-as-you-go billing required. Add payment method at openai.com

Error Handling

Error Codes

Prompt Validation

# Too short
{baseDir}/image-gen.js "Hi"
# ❌ ERROR: Prompt too short: 2 chars (min: 5)

# Too long (truncated here)
{baseDir}/image-gen.js "A" * 5000
# ❌ ERROR: Prompt too long: 5000 chars (max: 4000)

# Empty
{baseDir}/image-gen.js ""
# ❌ ERROR: Prompt is required

Examples by Use Case

1. Product Photography

# Basic product shot
{baseDir}/image-gen.js "Professional product photo of a black leather wallet on marble surface, soft lighting, commercial photography"

# Styled product
{baseDir}/image-gen.js "Premium coffee bag on wooden table, morning light, cafe ambiance, commercial photography, 35mm lens" --quality hd --size 1024x1792

# Multiple angles
{baseDir}/image-gen.js "Luxury watch floating on black background, studio lighting, product photography" --n 4

2. Concept Art

# Character design
{baseDir}/image-gen.js "Fantasy warrior in magical armor, digital art, concept art, trending on artstation" --custom-style digital_art

# Environment
{baseDir}/image-gen.js "Floating islands with waterfalls and glowing crystals, epic scale, concept art" --model dall-e-3 --size 1792x1024

# Mood board
{baseDir}/image-gen.js "Post-apocalyptic city ruins at golden hour, atmospheric, cinematic" --custom-style cinematic --n 4

3. Marketing Assets

# Social media images
{baseDir}/image-gen.js "Summer sale banner with beach scene, bright colors, text overlay ready" --size 1024x1024 --n 3

# App icons
{baseDir}/image-gen.js "Minimal app icon design, abstract geometric pattern, pastel colors" --model dall-e-2 --size 256x256

# Hero images
{baseDir}/image-gen.js "Diverse team collaborating in modern office, natural light, inclusive" --size 1792x1024 --quality hd

4. Creative Projects

# Book cover
{baseDir}/image-gen.js "Sci-fi book cover: astronaut discovering alien artifact, dramatic lighting, space opera" --size 1024x1792

# Album art
{baseDir}/image-gen.js "Surreal dreamscape with melting clocks and floating pianos, Dali style" --custom-style oil_painting

# Poster
{baseDir}/image-gen.js "Neon-lit Tokyo street at night, rain on pavement, cyberpunk" --custom-style neon --size 1024x1792

5. Architectural Visualization

# Exterior shot
{baseDir}/image-gen.js "Modern glass house in forest at dusk, warm interior lights, architectural photography" --custom-style photographic --size 1792x1024

# Interior design
{baseDir}/image-gen.js "Minimalist living room with natural wood furniture, large windows, interior design shot" --quality hd

6. Batch Processing Script

#!/bin/bash
# Generate images from prompts in file

while IFS= read -r prompt; do
  [ -z "$prompt" ] && continue
  [ "${prompt:0:1}" == "#" ] && continue  # Skip comments
  
  echo "Generating: $prompt"
  {baseDir}/image-gen.js "$prompt" --output-dir ./batch-output --n 2
  sleep 2  # Rate limit protection
done < prompts.txt

Technical Details

Architecture

1. Validate Prompt
   ├── Check length (5-4000 chars)
   ├── Apply style preset if specified
   └── Final prompt assembly

2. Check Limits
   ├── Rate limit (5/min, 1/sec)
   └── Skip if rate limit exceeded

3. API Request
   ├── HTTPS POST to OpenAI
   ├── Include model, size, quality, n
   └── Wait for base64 response

4. Process Response
   ├── Validate data received
   └── Track cost

5. Save Images
   ├── Decode base64
   ├── Generate filename
   ├── Write PNG
   └── Report size

6. Create Gallery (if n > 1)
   ├── Generate HTML
   └── Include metadata

API Request Structure

{
  "model": "dall-e-3",
  "prompt": "A serene mountain landscape at sunset",
  "n": 1,
  "size": "1024x1024",
  "quality": "standard",
  "response_format": "b64_json",
  "style": "vivid"
}

Response Handling

Error Responses:

• 429: Rate limited
• 400: Content policy violation
• 401: Invalid API key
• 500: OpenAI server error

Success Response:

{
  "created": 1705312345,
  "data": [
    {
      "b64_json": "base64_encoded_image_data...",
      "revised_prompt": "A serene mountain landscape at sunset..."
    }
  ]
}

Advanced Usage

Retry Logic

# If generation fails, retry with slight modification
for attempt in {1..3}; do
  {baseDir}/image-gen.js "A cat" --output-dir ./retry-test
  if [ $? -eq 0 ]; then break; fi
  echo "Retry $attempt..."
  sleep 5
done

Custom Style Templates

Create your own style by concatenating templates:

# Combine multiple style elements
STYLE="A photorealistic image of"
PROMPT="Lion resting in savanna at sunrise"
SUFFIX="golden hour lighting, wildlife photography, National Geographic style, detailed fur texture"
{baseDir}/image-gen.js "$STYLE $PROMPT, $SUFFIX"

Integration with Other Tools

# Generate and convert to JPEG
{baseDir}/image-gen.js "Abstract pattern" --output ./generated.png
mogrify -format jpg generated.png

# Generate and resize
{baseDir}/image-gen.js "Landscape" --size 1024x1024
sips -Z 512 generated.png

# Generate and optimize
{baseDir}/image-gen.js "Icon design"
pngquant --quality=65-80 generated.png -o optimized.png

Notes

• File Format: All images saved as PNG (lossless)
• Filenames: Generated based on prompt slug + timestamp
• Gallery: Auto-generated for batch generations
• Prompt Enhancement: OpenAI may revise your prompt for safety/quality
• Cost Tracking: Estimated based on OpenAI pricing (may vary)
• Timeout: 5 minute maximum wait
• Concurrent: Limited by rate limiting (sequential processing)
• Storage: Images saved to /tmp by default (may be cleaned up)
• Reproducibility: Same prompt may yield different results
• API Key: Must have access to DALL-E models
• Usage Limits: Based on OpenAI tier and budget
• Content Policy: Follows OpenAI's use policy strictly