WAN Image 2.2 Generation

Generate images using the WAN Image 2.2 workflow with dual LoRA support for enhanced control over image generation.

Overview

The WAN Image 2.2 workflow allows you to use two separate LoRA models simultaneously with different strength settings, providing fine-grained control over the generation process. This is particularly useful for combining different style or subject LoRAs to achieve specific artistic effects.

Important: When using model_type: "wan-image-2.2", both lora_low_name and lora_high_name parameters are required. The SDK will throw an error if either is missing.

Note: The workflow_type parameter is deprecated. Use model_type: "wan-image-2.2" instead.

Method Signature

client.images.generate(params: GenerateImageInput)

Parameters

Parameter	Type	Required	Description
prompt	string	Yes	The text prompt for image generation
width	number	Yes	Image width (64-2048)
height	number	Yes	Image height (64-2048)
model_id	string	Yes	UUID of the model to use
model_type	"wan-image-2.2"	Yes	Must be set to "wan-image-2.2"
lora_low_name	string	Yes	Path to the low-strength LoRA model (required for WAN Image 2.2)
lora_low_strength	number	No	Strength for low LoRA (0-2, default: 0.5)
lora_high_name	string	Yes	Path to the high-strength LoRA model (required for WAN Image 2.2)
lora_high_strength	number	No	Strength for high LoRA (0-2, default: 1.0)
batch_size	number	No	Number of images to generate (1-4, default: 1)
guidance	number	No	Guidance scale (0-100)
model_weight	number	No	Model weight (0-1)
webhook_url	string	No	URL for webhook notifications

Example Usage

Basic Generation with Dual LoRAs (Required)

// WAN Image 2.2 requires both LoRA parameters
const result = await client.images.generate({
  prompt: "A female model in vintage fashion, natural lighting",
  width: 1200,
  height: 1500,
  model_id: "your-model-uuid",
  model_type: "wan-image-2.2",
  lora_low_name: "wan2.2/wan2.2-lora-instagirl-2.0/Instagirlv2.0_lownoise.safetensors", // Required
  lora_low_strength: 0.5,
  lora_high_name: "wan2.2/wan2.2-lora-instagirl-2.0/Instagirlv2.0_hinoise.safetensors", // Required
  lora_high_strength: 1.0,
  batch_size: 1
});

console.log("Generated image:", result.image_url);

Error Handling for Missing LoRAs

The SDK will throw an error if you don't provide both LoRA names:

try {
  const result = await client.images.generate({
    prompt: "Professional portrait photography",
    width: 1024,
    height: 1024,
    model_id: "your-model-uuid",
    model_type: "wan-image-2.2"
    // Missing lora_low_name and lora_high_name - will throw error
  });
} catch (error) {
  console.error(error.message); 
  // "WAN Image 2.2 workflow requires both lora_low_name and lora_high_name parameters"
}

With Webhook for Progress Updates

const result = await client.images.generate({
  prompt: "Artistic fashion photography",
  width: 1200,
  height: 1500,
  model_id: "your-model-uuid",
  model_type: "wan-image-2.2",
  lora_low_name: "path/to/style-lora.safetensors",
  lora_low_strength: 0.4,
  lora_high_name: "path/to/subject-lora.safetensors",
  lora_high_strength: 0.9,
  webhook_url: "https://your-webhook-endpoint.com/progress"
});

Response

The method returns a GenerateImageOutput object:

{
  runId: string;
  status: string;
  progress: number;
  liveStatus: string;
  run_type: string;
  image_url?: string;
}

Notes

• The WAN Image 2.2 workflow uses a predefined negative prompt optimized for high-quality generation
• The dual LoRA system allows for complex style mixing not possible with single LoRA workflows
• LoRA strengths can be fine-tuned to balance between different styles or subjects
• The workflow is particularly effective for fashion, portrait, and artistic photography

Comparison with Other Workflows

Feature	Default	WAN	WAN Image 2.2
Single LoRA	✓	✓	✓
Dual LoRA	✗	✗	✓
Custom LoRA Strength	✓	✓	✓ (per LoRA)
Optimized Negative Prompt	✗	✗	✓
Batch Generation	✓	✓	✓

Error Handling

try {
  const result = await client.images.generate({
    // ... parameters
    workflow_type: "wan-image-2.2"
  });
} catch (error) {
  console.error("Generation failed:", error.message);
  // Handle specific error cases
}