The reverse video prompt generation operator is designed for video replication and storyboard recreation scenarios. The operator first understands source video facts. It then combines each storyboard segment's time range, visual description, and reference images to output executable video generation prompts by segment.

Operator introduction

Key features

Generates prompts segment by segment based on source video facts and storyboard time windows, avoiding content mixing across segments.

Preserves verified subject identity, spatial relationships, key action order, original dialogue, and voice performance.

Supports specifying the target visual style and target voice timbre without changing source video facts.

Supports binding reference images in storyboards and returning corresponding references in the generated results.

Supported regions

Beijing: cn-beijing

Shanghai: cn-shanghai

Guangzhou: cn-guangzhou

Operator performance

Sub-item	Detailed requirements
Submit RPM	200
Poll RPM	300

Input and output requirements

Input requirements

Sub-item	Detailed requirements
Supported input data modalities	Video Image Text
Input format: video	Duration: the original video duration range is 4 to 360 seconds. The storyboard segment duration is 4 to 15 seconds. A maximum of 24 inline storyboard segments are supported, and their time ranges must be monotonic and non-overlapping.
Input format: image	Each storyboard segment can have up to 9 reference images. Positions must start from 1 and increase continuously.
Input format: text	Size: the storyboard bundle manifest and each referenced JSON file cannot exceed 2 MiB.
Input path requirements	Provide the input data required by the operator through request parameters such as source_video_url and storyboard_bundle_url. Accessible HTTP/HTTPS URLs or TOS paths are supported. Public URLs that require a login state or additional Header authentication are not supported. Temporary URLs must remain valid during task execution. TOS objects must belong to the same primary account and region as the current LAS service and have read permissions.

Output requirements

Sub-item	Detailed requirements
Supported output data modalities	Text
Output format: text	Returns video prompts by storyboard segment in the input storyboard order, and returns reference images bound to each storyboard segment. Output language: output_language supports zh and en, and defaults to zh. Original dialogue remains in its original language.
Output method	After the task is completed, results are returned through data.items of the Poll API.

Operator pricing

Use your own ModelArk account to call the model or specify an endpoint. Specific costs are based on the actual model usage of that ModelArk account.

Notes and prerequisites

Enable ModelArk

This operator needs to call the ModelArk model that you have enabled. Bills are also issued under your own ModelArk model bill. Therefore, before using the operator, you need to enable the ModelArk model first and obtain the authentication information (ark_api_key) and model information (model ID and endpoint information) for ModelArk model calls.

General notes

Sub-item	Notes and prerequisites
Enable LAS	If you are a new Volcengine user and have not enabled the LAS product before, you can enable LAS first. Enabling only the LAS product without using LAS billing features does not incur fees. For the activation process, see Preparations. After activation, you can view the operator introduction document to learn about operator capabilities and getting started guides. For details, see LAS intelligent data processing operators.
Fees	Before calling the operator, you need to understand the model call fees when using the operator. For details, see Large model call billing.
Authentication (API Key)	Before calling the operator, you need to generate an API Key for operator calls. We recommend configuring the API Key as an environment variable for more secure operator calls. For details, see Obtain an API Key and configure it.
BaseURL	Before calling the operator, you need to obtain the BaseURL for operator calls based on the region where your current LAS service is located. It is used to configure path parameter values for operator calls. For details, see Obtain the Base URL. The examples below are for reference only. Replace the path value with the value for your corresponding region during actual calls.

REST API calls

Submit

API description

Submit an asynchronous task for reverse video prompt generation.

The request header uses Authorization: Bearer <LAS_API_KEY> for LAS service authentication. The authentication parameter value is the LAS API Key. The ark_api_key in the request body must use your own ModelArk API Key for authentication. All endpoints are optional. If omitted, the corresponding model ID is used. Lite is used for audio fact understanding. Standard or Pro is used for prompt expansion.

After success, the task ID is returned. You can query prompts by storyboard segment through Poll.

Request parameters

Response parameters

Request example

Response example

Error codes

HttpStatusCode	Error code	Error message	Description
400	Parameter.Invalid	The parameter is invalid.	The request parameters are invalid. For example, required ModelArk credentials are missing, the storyboard structure or time range does not meet constraints, or batch submission is attempted
401	Authorization.Missing	Missing Authorization.	LAS authentication is missing
401	ApiKey.Invalid	The api key is invalid.	The LAS API Key is invalid
500	Server.InternalError	General Internal Error.	Internal service error

Poll