fal-ai/omnigen-v2

Readme

OmniGen V2: Advanced AI Image Generation

Transform your creative vision into stunning images with OmniGen V2, a unified image generation model that handles multiple tasks within a single framework, from text-to-image generation to complex image editing.

Overview

OmniGen V2 delivers exceptional image generation capabilities through an intuitive API. Whether you're building a creative application, generating marketing assets, or exploring AI art, OmniGen V2 provides a unified approach that eliminates the need for additional modules like ControlNet or IP-Adapter.

Key Features

Transform text descriptions into high-quality images with advanced capabilities:

Unified Framework

• Text-to-image generation, image editing, and personalized generation in one model
• No need for additional modules or preprocessing steps
• Supports multi-modal prompts with text and image inputs

Versatile Applications

• Image editing with natural language instructions
• Personalized image generation using reference images
• Virtual try-on capabilities
• Multi-person generation
• Subject-driven generation

Developer Experience

• Simple REST API with comprehensive SDKs
• Support for up to 3 input images
• Extensive documentation and examples

Getting Started

Getting up and running with OmniGen V2 takes just a few minutes. Here's how:

1. Install the SDK for your platform:

JavaScript/TypeScript:

bash
npm install --save @fal-ai/client

Python:

bash
pip install fal-client

2. Configure your credentials:

JavaScript:

javascript
import { fal } from "@fal-ai/client";

fal.config({
  credentials: "YOUR_FAL_KEY"
});

Python:

python
import fal_client
import os

os.environ["FAL_KEY"] = "YOUR_FAL_KEY"

3. Generate your first image:

JavaScript:

javascript
const result = await fal.subscribe("fal-ai/omnigen-v2", {
  input: {
    prompt: "A serene mountain landscape at sunset, photorealistic style"
  }
});

Python:

python
result = fal_client.subscribe("fal-ai/omnigen-v2", {
    "prompt": "A serene mountain landscape at sunset, photorealistic style"
})

Advanced Usage

Image Editing with Reference Images

javascript
const result = await fal.subscribe("fal-ai/omnigen-v2", {
  input: {
    prompt: "Make the dress blue",
    input_image_urls: ["https://your-image-url.jpg"]
  }
});

Multi-Image Generation

javascript
const result = await fal.subscribe("fal-ai/omnigen-v2", {
  input: {
    prompt: "Add the bird from image 1 to the desk in image 2",
    input_image_urls: [
      "https://bird-image.jpg",
      "https://desk-image.jpg"
    ]
  }
});

API Parameters

Input Parameters:

• prompt (required): Text description for generation or editing
• input_image_urls: Array of up to 3 image URLs for editing/reference
• image_size: Output size (default: "square_hd")
  • Options: square_hd, square, portrait_4_3, portrait_16_9, landscape_4_3, landscape_16_9
• num_inference_steps: Number of denoising steps (default: 50)
• text_guidance_scale: Controls prompt adherence (default: 5)
• image_guidance_scale: Controls reference image influence (default: 2)
• negative_prompt: Elements to avoid in generation
• scheduler: Sampling method (default: "euler")
• num_images: Number of images to generate (default: 1)
• enable_safety_checker: Enable content filtering (default: true)
• output_format: Image format (default: "jpeg")

Queue-Based Processing

For asynchronous generation:

javascript
// Submit request
const { request_id } = await fal.queue.submit("fal-ai/omnigen-v2", {
  input: {
    prompt: "Your prompt here"
  },
  webhookUrl: "https://optional.webhook.url/for/results"
});

// Check status
const status = await fal.queue.status("fal-ai/omnigen-v2", {
  requestId: request_id,
  logs: true
});

// Get result
const result = await fal.queue.result("fal-ai/omnigen-v2", {
  requestId: request_id
});

Best Practices

Maximize the quality of your generations with these proven approaches:

Write Effective Prompts

• Use specific language for multi-image tasks (e.g., "Add the X from image 1 to Y in image 2")
• Be descriptive about style, mood, and technical details
• Include negative prompts to avoid unwanted elements

Optimize Parameters

• Adjust text_guidance_scale (1-20) for prompt adherence
• Use image_guidance_scale (1-10) to control reference image influence
• Higher inference steps (50-100) generally improve quality

Handle Results Properly

• Implement proper error handling for API responses
• Use the file storage API for uploading local images
• Monitor usage through the fal.ai dashboard

File Upload Support

Upload local images for editing or reference:

javascript
import { fal } from "@fal-ai/client";

const file = new File([imageData], "image.jpg", { type: "image/jpeg" });
const url = await fal.storage.upload(file);

const result = await fal.subscribe("fal-ai/omnigen-v2", {
  input: {
    prompt: "Edit this image",
    input_image_urls: [url]
  }
});

Technical Specifications

Model Architecture

• Based on unified diffusion technology
• Simplified architecture without additional text encoders
• Supports arbitrarily interleaved text and image inputs

Performance

• Fast inference times
• Supports various output resolutions and aspect ratios
• Automatic safety checking

Integration Options

• Native SDKs for JavaScript and Python
• Direct HTTP API access
• Webhook support for async processing

Pricing and Usage

• Cost: $0.15 per megapixel
• Transparent, usage-based pricing
• No subscription necessary
• No hidden fees or minimum commitments

View detailed pricing or contact sales for enterprise solutions.

Support and Resources

We're here to help you succeed with OmniGen V2:

Documentation

• API Documentation
• General fal.ai Documentation
• Code examples in multiple languages

Community and Support

• Active developer community
• Direct email support
• Regular model updates

Get started with OmniGen V2 today and experience the power of unified image generation. Create your API key at fal.ai to begin generating amazing images with our API.