Image Generation

Overview

LlmTornado provides powerful image generation capabilities through multiple AI providers including OpenAI (DALL-E), Google (Imagen), and X.AI (Grok). You can generate, edit, and manipulate images using natural language prompts. The library supports various image formats, qualities, and styles to suit different use cases.

Quick Start

Here's a simple image generation example:

csharp

using LlmTornado;
using LlmTornado.Images;
using LlmTornado.Images.Models;

TornadoApi api = new TornadoApi("your-api-key");

// Generate an image
ImageGenerationResult? result = await api.ImageGenerations.CreateImage(
    new ImageGenerationRequest("a cute cat sitting on a sunny windowsill")
    {
        Quality = TornadoImageQualities.Hd,
        ResponseFormat = TornadoImageResponseFormats.Url,
        Model = ImageModel.OpenAi.Dalle.V3
    });

Console.WriteLine(result?.Data?[0].Url);

Prerequisites

Before using image generation features, ensure you have:

The LlmTornado package installed
A valid API key with image generation access
Understanding of image generation concepts and prompts
Sufficient API credits (image generation can be expensive)

Detailed Explanation

Image Generation Models

LlmTornado supports multiple image generation models:

OpenAI Models

DALL-E 3 - High-quality, detailed images with excellent prompt following
DALL-E 2 - Fast generation with good quality
GPT-1 (Image) - Latest generation model with advanced features

Google Models

Imagen 3 - Google's latest image generation model
Imagen 4 Preview - Preview of next-generation features

X.AI Models

Grok 2 - X.AI's image generation model

Response Formats

Images can be returned in two formats:

URL - Direct link to the generated image (expires after a period)
Base64 - Encoded image data for immediate use or storage

Quality Options

Control the output quality:

Standard - Fast generation, good quality
HD - Higher quality, more detailed images
Medium - Balanced quality and speed

Basic Usage

Generate with DALL-E 3

csharp

ImageGenerationResult? result = await api.ImageGenerations.CreateImage(
    new ImageGenerationRequest("a cute cat", 
        quality: TornadoImageQualities.Hd, 
        responseFormat: TornadoImageResponseFormats.Url, 
        model: ImageModel.OpenAi.Dalle.V3));

Console.WriteLine($"Image URL: {result?.Data?[0].Url}");

Generate with Base64 Response

csharp

ImageGenerationResult? result = await api.ImageGenerations.CreateImage(
    new ImageGenerationRequest("a cute cat", 
        quality: TornadoImageQualities.Hd, 
        responseFormat: TornadoImageResponseFormats.Base64, 
        model: ImageModel.OpenAi.Dalle.V3));

if (result?.Data?[0].Base64 != null)
{
    byte[] imageBytes = Convert.FromBase64String(result.Data[0].Base64);
    await File.WriteAllBytesAsync("generated_image.jpg", imageBytes);
}

Generate with GPT-1

csharp

ImageGenerationResult? result = await api.ImageGenerations.CreateImage(
    new ImageGenerationRequest("a cute cat", 
        quality: TornadoImageQualities.Medium, 
        model: ImageModel.OpenAi.Gpt.V1)
    {
        Background = ImageBackgroundTypes.Transparent,
        Moderation = ImageModerationTypes.Low
    });

byte[] imageBytes = Convert.FromBase64String(result.Data[0].Base64);
await File.WriteAllBytesAsync("cat_transparent.png", imageBytes);

Generate with Google Imagen

csharp

ImageGenerationResult? result = await api.ImageGenerations.CreateImage(
    new ImageGenerationRequest("a cute cat", 
        responseFormat: TornadoImageResponseFormats.Base64, 
        model: ImageModel.Google.Imagen.V3Generate002)
    {
        VendorExtensions = new ImageGenerationRequestVendorExtensions(
            new ImageGenerationRequestGoogleExtensions
            {
                MimeType = ImageGenerationRequestGoogleExtensionsMimeTypes.Jpeg,
                CompressionQuality = 90
            })
    });

byte[] imageBytes = Convert.FromBase64String(result.Data[0].Base64);
await File.WriteAllBytesAsync("cat_imagen.jpg", imageBytes);

Generate with Grok 2

csharp

ImageGenerationResult? result = await api.ImageGenerations.CreateImage(
    new ImageGenerationRequest
    {
        Prompt = "a cute cat in a cyberpunk setting",
        ResponseFormat = TornadoImageResponseFormats.Base64,
        Model = ImageModel.XAi.Grok.V2241212
    });

byte[] imageBytes = Convert.FromBase64String(result.Data[0].Base64);
await File.WriteAllBytesAsync("cyberpunk_cat.jpg", imageBytes);

Advanced Usage

Image Editing

Edit existing images with prompts:

csharp

// First, generate an image
ImageGenerationResult? original = await api.ImageGenerations.CreateImage(
    new ImageGenerationRequest("a cute cat", 
        quality: TornadoImageQualities.Medium, 
        model: ImageModel.OpenAi.Gpt.V1)
    {
        Background = ImageBackgroundTypes.Transparent,
        Moderation = ImageModerationTypes.Low
    });

// Then edit it
ImageGenerationResult? edited = await api.ImageEdit.EditImage(
    new ImageEditRequest("make this cat look more dangerous")
    {
        Quality = TornadoImageQualities.Medium,
        Model = ImageModel.OpenAi.Gpt.V1,
        Image = new TornadoInputFile(original.Data[0].Base64, "image/png")
    });

byte[] editedBytes = Convert.FromBase64String(edited.Data[0].Base64);
await File.WriteAllBytesAsync("dangerous_cat.png", editedBytes);

Advanced Imagen 4 Generation

Use Google's Imagen 4 preview with advanced settings:

csharp

ImageGenerationResult? result = await api.ImageGenerations.CreateImage(
    new ImageGenerationRequest("a cute cat", 
        responseFormat: TornadoImageResponseFormats.Base64, 
        model: ImageModel.Google.ImagenPreview.V4Preview250606)
    {
        VendorExtensions = new ImageGenerationRequestVendorExtensions(
            new ImageGenerationRequestGoogleExtensions
            {
                MimeType = ImageGenerationRequestGoogleExtensionsMimeTypes.Jpeg,
                CompressionQuality = 90,
                AspectRatio = "16:9", // Widescreen format
                NegativePrompt = "blurry, low quality"
            })
    });

byte[] imageBytes = Convert.FromBase64String(result.Data[0].Base64);
await File.WriteAllBytesAsync("cat_widescreen.jpg", imageBytes);

Batch Image Generation

Generate multiple images at once:

csharp

string[] prompts = [
    "a cute cat in a garden",
    "a playful kitten with a ball of yarn",
    "an elegant cat on a velvet cushion"
];

List<ImageGenerationResult?> results = [];

foreach (string prompt in prompts)
{
    ImageGenerationResult? result = await api.ImageGenerations.CreateImage(
        new ImageGenerationRequest(prompt, 
            quality: TornadoImageQualities.Hd, 
            model: ImageModel.OpenAi.Dalle.V3));
    
    results.Add(result);
    Console.WriteLine($"Generated: {prompt} -> {result?.Data?[0].Url}");
    
    // Respect rate limits
    await Task.Delay(1000);
}

Saving and Displaying Images

Helper method to save generated images:

csharp

async Task SaveGeneratedImage(ImageGenerationResult? result, string filename)
{
    if (result?.Data?[0] != null)
    {
        if (!string.IsNullOrEmpty(result.Data[0].Base64))
        {
            byte[] imageBytes = Convert.FromBase64String(result.Data[0].Base64);
            await File.WriteAllBytesAsync(filename, imageBytes);
            Console.WriteLine($"Image saved to {filename}");
        }
        else if (!string.IsNullOrEmpty(result.Data[0].Url))
        {
            using HttpClient client = new HttpClient();
            byte[] imageBytes = await client.GetByteArrayAsync(result.Data[0].Url);
            await File.WriteAllBytesAsync(filename, imageBytes);
            Console.WriteLine($"Image downloaded and saved to {filename}");
        }
    }
}

// Usage
ImageGenerationResult? result = await api.ImageGenerations.CreateImage(
    new ImageGenerationRequest("a sunset over mountains"));
await SaveGeneratedImage(result, "sunset.jpg");

Error Handling

Robust error handling for image generation:

csharp

try
{
    ImageGenerationResult? result = await api.ImageGenerations.CreateImage(
        new ImageGenerationRequest("a cute cat")
        {
            Quality = TornadoImageQualities.Hd,
            Model = ImageModel.OpenAi.Dalle.V3
        });

    if (result?.Data?.Count > 0)
    {
        Console.WriteLine("Image generated successfully!");
    }
    else
    {
        Console.WriteLine("No image data returned");
    }
}
catch (Exception ex)
{
    Console.WriteLine($"Error generating image: {ex.Message}");
    // Handle specific errors
    if (ex.Message.Contains("rate limit"))
    {
        Console.WriteLine("Rate limit exceeded. Please wait and try again.");
    }
    else if (ex.Message.Contains("content policy"))
    {
        Console.WriteLine("Prompt violates content policy. Please modify your prompt.");
    }
}

Best Practices

1. Write Effective Prompts

Be specific and descriptive
Include style, mood, and composition details
Mention lighting, colors, and perspective
Use negative prompts to avoid unwanted elements (where supported)

2. Manage Costs

Use standard quality for testing
Use HD quality only when necessary
Cache and reuse generated images
Consider cheaper models for simple images

3. Handle Rate Limits

Implement delays between requests
Use exponential backoff for retries
Monitor your API usage
Queue requests for batch processing

4. Store Images Efficiently

Save base64 images to disk/database immediately
Download URL images before they expire
Compress images for storage if quality allows
Use appropriate image formats (JPEG vs PNG)

5. Content Moderation

Validate prompts before sending
Handle content policy violations gracefully
Use moderation settings appropriately
Implement user prompt filtering

Common Issues

Content Policy Violation

Issue: Prompt rejected by content filter
Solution: Modify prompt to remove problematic content
Prevention: Validate prompts against policy guidelines

Rate Limit Exceeded

Issue: Too many generation requests
Solution: Implement rate limiting and queuing
Prevention: Monitor usage and add delays between requests

Image Quality Issues

Issue: Generated images don't match expectations
Solution: Refine prompts with more specific details
Prevention: Test prompts and iterate on descriptions

Expired URLs

Issue: Image URLs no longer accessible
Solution: Download and store images immediately
Prevention: Use base64 format or download URLs promptly

API Reference

ImageGenerationRequest

string Prompt - Description of image to generate
ImageModel Model - Model to use for generation
TornadoImageQualities Quality - Output quality level
TornadoImageResponseFormats ResponseFormat - URL or Base64
ImageBackgroundTypes Background - Background type (transparent, opaque)
ImageModerationTypes Moderation - Content moderation level
ImageGenerationRequestVendorExtensions VendorExtensions - Provider-specific options

ImageGenerationResult

List<ImageData> Data - Generated image data
int Created - Timestamp of creation
string Model - Model used

ImageData

string Url - Direct image URL (if requested)
string Base64 - Base64 encoded image (if requested)
string RevisedPrompt - AI-revised prompt (some models)

ImageModel

OpenAi.Dalle.V3 - DALL-E 3 model
OpenAi.Dalle.V2 - DALL-E 2 model
OpenAi.Gpt.V1 - GPT image model
Google.Imagen.V3Generate002 - Imagen 3
Google.ImagenPreview.V4Preview250606 - Imagen 4 Preview
XAi.Grok.V2241212 - Grok 2 image model

Vision - Image understanding and analysis
Files - File management
Chat Basics - Core chat functionality

Image Generation ​

Overview ​

Quick Start ​

Prerequisites ​

Detailed Explanation ​

Image Generation Models ​

OpenAI Models ​

Google Models ​

X.AI Models ​

Response Formats ​

Quality Options ​

Basic Usage ​

Generate with DALL-E 3 ​

Generate with Base64 Response ​

Generate with GPT-1 ​

Generate with Google Imagen ​

Generate with Grok 2 ​

Advanced Usage ​

Image Editing ​

Advanced Imagen 4 Generation ​

Batch Image Generation ​

Saving and Displaying Images ​

Error Handling ​

Best Practices ​

1. Write Effective Prompts ​

2. Manage Costs ​

3. Handle Rate Limits ​

4. Store Images Efficiently ​

5. Content Moderation ​

Common Issues ​

Content Policy Violation ​

Rate Limit Exceeded ​

Image Quality Issues ​

Expired URLs ​

API Reference ​

ImageGenerationRequest ​

ImageGenerationResult ​

ImageData ​

ImageModel ​

Related Topics ​

Image Generation

Overview

Quick Start

Prerequisites

Detailed Explanation

Image Generation Models

OpenAI Models

Google Models

X.AI Models

Response Formats

Quality Options

Basic Usage

Generate with DALL-E 3

Generate with Base64 Response

Generate with GPT-1

Generate with Google Imagen

Generate with Grok 2

Advanced Usage

Image Editing

Advanced Imagen 4 Generation

Batch Image Generation

Saving and Displaying Images

Error Handling

Best Practices

1. Write Effective Prompts

2. Manage Costs

3. Handle Rate Limits

4. Store Images Efficiently

5. Content Moderation

Common Issues

Content Policy Violation

Rate Limit Exceeded

Image Quality Issues

Expired URLs

API Reference

ImageGenerationRequest

ImageGenerationResult

ImageData

ImageModel

Related Topics