PixVerse Platform Docs(for enterprise)
  1. Fusion(Reference to Video)
PixVerse Platform Docs(for enterprise)
  • Model & Pricing
  • Multi-transition
    • How to use multi-transtion
    • Multi-trasition generate
      POST
  • Previous updates
    • Sound_effect
      • How to use Sound effects?
      • Generate Sound Effects
    • Lip sync
      • How to use Lip Sync?
      • Tts speaker list
      • Generate lip sync video
      • Get tts speaker list
    • Fusion(Reference to Video)
      • How to use fusion (reference to video)?
      • Generate Fusion(Reference to Video) video
        POST
    • Template for multiple-images
      • How to use multiple-image templates?
    • Extend
      • How to use Extend?
      • Generate Extend video
    • Restyle
      • How to use restyle?
      • Restyle generation
      • Restyle get list
  • Upload media (video&audio)
    POST
  1. Fusion(Reference to Video)

How to use fusion (reference to video)?

The Fusion(Reference to Video) feature allows you to quickly generate a video based on multiple reference images.
These reference images can include subjects and backgrounds, and you can use prompts to help compose them naturally into a coherent scene.
Fusion is suitable for various scenarios such as e-commerce, brand advertising, and narrative videos.

Prerequisites#

Before you begin, make sure you have:
A valid PixVerse API key
A unique Ai-Trace-Id for each API request
An active subscription
The img_id(s) of the reference images you want to use in Fusion

step-step#

Step 1. Prepare Your Image#

Use high-quality and clear images. Higher resolution is recommended for better results.
For subject images, choose images where the main subject is clearly isolated and prominent.
For background images, select images with a clean and well-defined background.

Step 2. Prepare Your upload image API Request and uplaod image to get img_id#

Construct your API request with the appropriate parameters:

curl --location --request POST 'https://app-api.pixverse.ai/openapi/v2/image/upload' \
--header 'API-KEY:your-api-key' \
--header 'Ai-trace-id: your-Ai-trace-id' \
--form 'image=@""'

Step 3. Prepare your Fusion(reference-to-video ) API#

Parameter Details

Parameter Details#

Parameter NameRequiredTypeDescription
image_referencesRequiredarrayArray of reference images (1–3 items), including subjects and/or backgrounds
promptRequiredstringTo accurately describe the scene, use @ref_name in the prompt.
1. A space must follow @ref_name, e.g., @cat plays
2. The @name used in the prompt must exactly match the ref_name in image_references.
modelRequiredstringModel version, supported in model v4.5 and above
durationRequiredintVideo duration in seconds, supported 5, 8
qualityRequiredstringVideo resolution
aspect_ratioRequiredstringVideo aspect ratio
seedOptionalintRandom seed (used for result reproduction)

Notes & Considerations:#

image_references must be a non-empty array (1–3 items), and each item must be an object containing valid fields: "img_id", "ref_name", and "type".
"type" must be a valid value — please refer to the API documentation for supported options.
"ref_name" must be unique within the array.
The Fusion feature is only supported with model v4.5 and above.
In the prompt, when referencing a ref_name, you must:
Prefix it with @
Leave a space after it (e.g., @dog plays)
Ensure it exactly matches a ref_name defined in image_references.

Step 4.Handle the API Response#

The API returns a JSON response with a video_id:
{
"ErrCode": 0,
"ErrMsg": "success",
"Resp": {
"video_id": 0
}
}

Step 5.Check Generation Status#

After creating the task, you will receive a video_id
Query periodically Get Video Generation Status API using this video_id
The status will change from 5 to 1 when processing is complete
{
"ErrCode": 0,
"ErrMsg": "string",
"Resp": {
"create_time": "string",
"id": 0,
"modify_time": "string",
"negative_prompt": "string",
"outputHeight": 0,
"outputWidth": 0,
"prompt": "string",
"resolution_ratio": 0,
"seed": 0,
"size": 0,
"status": 5,
"style": "string",
"url": "string"
}
}

Step 6. Download the Generated Video#

You can access a generated video with "url"
{
"ErrCode": 0,
"ErrMsg": "string",
"Resp": {
"create_time": "string",
"id": 0,
"modify_time": "string",
"negative_prompt": "string",
"outputHeight": 0,
"outputWidth": 0,
"prompt": "string",
"resolution_ratio": 0,
"seed": 0,
"size": 0,
"status": 1,
"style": "string",
"url": "string"
}
}

Trobule shooting#

Common issue#

1.
Your video is stuck in "Generating" status and hasn't completed after a long wait.
Please check if you're using the same AI-trace-ID for every request. This is the most common cause of this issue.
2.
Status codes: 1: Generation successful; 5: Waiting for generation; 7: Content moderation failure; 8: Generation failed;
If you encounter status code 7, it means your generated video was filtered by our content moderation system. Please modify your parameters and try again. Any credits used for filtered videos will be automatically refunded to your account.

Common error codes#

400/500 status : Incorrect code
400013 : Invalid binding request: incorrect parameter type or value
400017 : Invalid parameter
500044 : Reached the limit for concurrent generations.
Modified at 2025-07-22 09:44:11
Previous
Get tts speaker list
Next
Generate Fusion(Reference to Video) video
Built with