Image Generation

Generate images using AI models in Bud Runtime

Overview

Bud Runtime supports image generation through various AI models, providing a unified API for creating images from text prompts.

Supported Models

• Stable Diffusion (SD 1.5, SDXL)
• Custom fine-tuned models
• LoRA-enhanced models

Basic Usage

Generate an Image

curl -X POST http://your-gateway/v1/images/generations \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "A serene landscape with mountains and a lake",
    "n": 1,
    "size": "1024x1024"
  }'

Response Format

{
  "created": 1234567890,
  "data": [
    {
      "url": "https://your-storage/image-id.png"
    }
  ]
}

Advanced Features

Using LoRA Models

Combine base models with LoRA adapters for specialized image generation:

{
  "prompt": "Portrait in anime style",
  "model": "stable-diffusion-xl",
  "lora": "anime-style-v2",
  "lora_weight": 0.8
}

Batch Generation

Generate multiple images in a single request:

{
  "prompt": "Futuristic cityscape",
  "n": 4,
  "size": "512x512",
  "seed": 42
}

Performance Optimization

• Use appropriate image sizes for your use case
• Enable caching for frequently used prompts
• Consider batch processing for multiple generations

Best Practices

1. Prompt Engineering: Be specific and descriptive in your prompts
2. Resource Management: Monitor GPU usage when generating large images
3. Quality Settings: Balance between generation speed and image quality

For more details on model configuration, see Multi-LoRA Support.