Automatic1111 Provider for Vercel AI SDK

A TypeScript provider for the Vercel AI SDK that enables image generation using AUTOMATIC1111 Stable Diffusion WebUI.

Features

• 🎨 Image Generation: Generate images using Stable Diffusion models via AUTOMATIC1111's API
• 🎯 Type Safety: Full TypeScript support with proper type definitions
• 🚀 Modern API: Built on the latest Vercel AI SDK v2 specification
• 🔄 Error Handling: Comprehensive error handling with detailed error messages

Installation

npm install automatic1111-provider

Prerequisites

Before using this provider, you need to have AUTOMATIC1111 running with the API enabled:

1. Install AUTOMATIC1111: Follow the official installation guide
2. Enable API: Start AUTOMATIC1111 with the --api flag:

   # Windows
   ./webui.bat --api
   
   # Linux/Mac
   ./webui.sh --api

3. Verify Setup: Your AUTOMATIC1111 instance should be accessible at http://127.0.0.1:7860 by default

Quick Start

import { experimental_generateImage as generateImage } from 'ai';
import { createAutomatic1111 } from 'automatic1111-provider';

// Create the provider instance
const automatic1111 = createAutomatic1111({
  baseURL: 'http://127.0.0.1:7860', // Your AUTOMATIC1111 instance
});

// Generate an image
const { images } = await generateImage({
  model: automatic1111.image('v1-5-pruned-emaonly'), // Your model name
  prompt: 'A beautiful sunset over mountains',
  n: 1,
  size: '512x512',
});

console.log('Generated image:', images[0]);

Configuration

Provider Settings

import { createAutomatic1111 } from 'automatic1111-provider';

const automatic1111 = createAutomatic1111({
  baseURL: 'http://127.0.0.1:7860', // AUTOMATIC1111 API endpoint
});

Provider Options

You can customize the image generation with AUTOMATIC1111-specific options:

const { images } = await generateImage({
  model: automatic1111.image('your-model-name'),
  prompt: 'A cyberpunk cityscape at night',
  providerOptions: {
    automatic1111: {
      negative_prompt: 'blurry, low quality, distorted',
      steps: 30,
      cfg_scale: 7.5,
      sampler_name: 'euler a',
      denoising_strength: 0.7,
      styles: ['anime', 'high quality'],
      check_model_exists: true
    }
  }
});

Available Options

Usage Examples

Basic Image Generation

import { experimental_generateImage as generateImage } from 'ai';
import { automatic1111 } from 'automatic1111-provider';

const result = await generateImage({
  model: automatic1111.image('sd_xl_base_1.0'),
  prompt: 'A majestic dragon flying over a medieval castle',
  size: '1024x1024',
});

Advanced Configuration

const result = await generateImage({
  model: automatic1111.image('realistic-vision-v4'),
  prompt: 'Portrait of a wise old wizard with a long beard',
  n: 2, // Generate 2 images
  seed: 12345, // Fixed seed for reproducibility
  providerOptions: {
    automatic1111: {
      negative_prompt: 'blurry, ugly, deformed, low quality',
      steps: 40,
      cfg_scale: 8.5,
      sampler_name: 'DPM++ SDE Karras',
      styles: ['photorealistic', 'detailed'],
    }
  }
});

Troubleshooting

Common Issues

1. Connection Refused

   • Ensure AUTOMATIC1111 is running with --api flag
   • Check if the baseURL is correct (default: http://127.0.0.1:7860)

2. Model Not Found

   • Verify the model name matches exactly what's in AUTOMATIC1111
   • Check that the model is properly loaded in the WebUI

3. Out of Memory

   • Reduce image size (e.g., from 1024x1024 to 512x512)
   • Lower the number of steps
   • Use --medvram or --lowvram flags when starting AUTOMATIC1111

4. Slow Generation

   • Reduce the number of steps
   • Use faster samplers like "Euler a" or "DPM++ 2M"
   • Enable xFormers in AUTOMATIC1111 settings

Performance Tips

• GPU Memory: AUTOMATIC1111 requires significant GPU memory. 8GB+ VRAM recommended
• Batch Size: Generate multiple images in a single call rather than multiple separate calls

API Reference

createAutomatic1111(options?)

Creates a new Automatic1111 provider instance.

Parameters

• options.baseURL (string?): API endpoint URL (default: 'http://127.0.0.1:7860')
• options.headers (object?): Custom HTTP headers
• options.apiKey (string?): API key (Not needed)

Returns

An Automatic1111Provider instance with image generation capabilities.

automatic1111

Pre-configured provider instance using default settings.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

MIT License - see the LICENSE file for details.

Related

• Vercel AI SDK - The AI SDK this provider is built for
• AUTOMATIC1111 - Stable Diffusion WebUI
• Stable Diffusion - The AI model powering image generation