Image Generation Skill

Generate images from text descriptions using AI.

When to Use

✅ USE this skill when:

• User asks to "create an image of..."
• "Generate a picture showing..."
• Need illustrations for content
• Visual brainstorming or concept art

When NOT to Use

❌ DON'T use this skill when:

• Editing existing images → use image editing tools
• Analyzing image content → use vision/OCR tools
• Creating diagrams/charts → use charting libraries
• Text-to-speech → use TTS services

Setup

Requires OPENAI_API_KEY environment variable.

export OPENAI_API_KEY="sk-..."

API Usage

Basic Image Generation

curl -X POST "https://api.openai.com/v1/images/generations" \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "dall-e-3",
    "prompt": "A sunset over mountains",
    "n": 1,
    "size": "1024x1024"
  }'

Response Format

{
  "created": 1234567890,
  "data": [
    {
      "url": "https://...",
      "revised_prompt": "..."
    }
  ]
}

Size Options

• 1024x1024 (default)
• 1024x1792 (portrait)
• 1792x1024 (landscape)

Quality Options

• standard (default, faster)
• hd (higher quality, slower)

Style Options

• vivid (default, dramatic)
• natural (photorealistic)

Advanced Options

With Custom Parameters

curl -X POST "https://api.openai.com/v1/images/generations" \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "dall-e-3",
    "prompt": "A cyberpunk city at night",
    "n": 1,
    "size": "1792x1024",
    "quality": "hd",
    "style": "vivid"
  }'

Multiple Variations (DALL-E 2)

# First generate the base image
BASE_RESPONSE=$(curl -X POST "https://api.openai.com/v1/images/generations" \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "dall-e-2",
    "prompt": "A cat",
    "n": 1,
    "size": "1024x1024"
  }')

# Then create variations (requires image URL from first response)
curl -X POST "https://api.openai.com/v1/images/variations" \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -F "image=@image.png" \
  -F "n=4" \
  -F "size=1024x1024"

Node.js Implementation

const fetch = require('node-fetch');

async function generateImage(prompt, options = {}) {
  const response = await fetch('https://api.openai.com/v1/images/generations', {
    method: 'POST',
    headers: {
      'Authorization': `Bearer ${process.env.OPENAI_API_KEY}`,
      'Content-Type': 'application/json'
    },
    body: JSON.stringify({
      model: options.model || 'dall-e-3',
      prompt: prompt,
      n: options.count || 1,
      size: options.size || '1024x1024',
      quality: options.quality || 'standard',
      style: options.style || 'vivid'
    })
  });

  const data = await response.json();
  
  if (!response.ok) {
    throw new Error(data.error?.message || 'Image generation failed');
  }

  return {
    urls: data.data.map(img => img.url),
    revisedPrompts: data.data.map(img => img.revised_prompt),
    created: data.created
  };
}

// Download image helper
async function downloadImage(url, outputPath) {
  const fs = require('fs');
  const response = await fetch(url);
  const buffer = await response.buffer();
  fs.writeFileSync(outputPath, buffer);
  return outputPath;
}

// Usage
const result = await generateImage('A futuristic city with flying cars', {
  size: '1792x1024',
  quality: 'hd',
  style: 'vivid'
});

console.log('Generated:', result.urls[0]);
await downloadImage(result.urls[0], '/tmp/generated.png');

Prompt Tips

Good Prompts

• Be specific: "A red apple on a wooden table" vs "apple"
• Include style: "photorealistic", "oil painting", "digital art"
• Set the mood: "sunset lighting", "dramatic shadows"
• Define composition: "close-up", "wide angle", "portrait"

Prompt Template

[subject] + [action/state] + [environment] + [lighting] + [style]

Example: "A wise owl (subject) perched on a branch (action) 
in an enchanted forest (environment) with moonlight filtering 
through leaves (lighting), digital painting style (style)"

Avoid

• Too many subjects (keep it focused)
• Contradictory descriptions
• Overly complex scenes
• Text within images (DALL-E struggles with text)

Error Handling

async function safeGenerateImage(prompt) {
  try {
    const result = await generateImage(prompt);
    return { success: true, ...result };
  } catch (error) {
    if (error.message.includes('content_policy')) {
      return { success: false, error: 'Content violated policy' };
    }
    if (error.message.includes('rate_limit')) {
      return { success: false, error: 'Rate limit exceeded' };
    }
    return { success: false, error: error.message };
  }
}

Cost Reference

• DALL-E 3 Standard: ~$0.040/image (1024x1024)
• DALL-E 3 HD: ~$0.080/image (1024x1024)
• DALL-E 2: ~$0.020/image (1024x1024)

Quick Response Template

"Generate an image of [X]"

const result = await generateImage(prompt, { size: '1024x1024' });
return `🎨 **Image Generated**

**Prompt:** ${prompt}
**Revised:** ${result.revisedPrompts[0]}

![Generated Image](${result.urls[0]})

[Download](${result.urls[0]})
`;

Notes

• DALL-E 3 takes ~10-30 seconds per image
• Images expire after ~1 hour — download immediately
• Content policy restrictions apply (no violence, celebrities, etc.)
• 4096 characters max prompt length