Operations

Description

Operations represent asynchronous tasks in the system. They can be created, monitored, and updated. The system supports various operation types like image generation, uploading, upscaling, and more.

Methods

create

Contains methods for creating different types of operations:

• create.describeProduct() - Creates a product description operation
• create.generate() - Creates an image generation operation
• create.imagine() - Creates an imagination operation
• create.proposePrompt() - Creates a prompt proposal operation
• create.replaceArea() - Creates an area replacement operation
• create.select() - Creates a selection operation
• create.upload() - Creates an upload operation
• create.upscale() - Creates an upscale operation

Each creation method accepts the following props:

Name	Type	Required	Description
input	any	Yes	The input data for the operation
metadata	any	No	Additional metadata for the operation

Returns

Promise<Operation<Input, Output>> - The created operation object

Example

const operation = await client.operations.create.generate({
    input: {
        aspectRatio: '1:1',
        productImageId: '123',
        prompt: 'Woman in a red dress on a red background',
        quality: 'high',
        seed: sdk.utils.seed(),
        version: 1,
    },
});

get

Retrieves a single operation by its ID.

Props

Name	Type	Required	Description
id	string	Yes	The operation's ID

Returns

Promise<Operation<any, any> | undefined> - The operation object if found

Example

const operation = await client.operations.get({ id: '123' });

getMany

Retrieves multiple operations by their IDs.

Props

Name	Type	Required	Description
ids	string[]	Yes	Array of operation IDs to get

Returns

Promise<Operation<any, any>[]> - Array of operation objects

Example

const operations = await client.operations.getMany({ ids: ['123', '456'] });

list

Lists operations with optional pagination.

Props

Name	Type	Required	Description
lastId	string	No	ID of the last item for pagination
limit	number	No	Maximum number of items to return

Returns

Promise<Operation<any, any>[]> - Array of operation objects

Example

const operations = await client.operations.list({ lastId: '123', limit: 10 });

update

Updates an existing operation.

Props

Name	Type	Required	Description
id	string	Yes	The operation's ID
metadata	any	No	Updated metadata for operation

Returns

Promise<void>

Example

await client.operations.update({ id: '123', metadata: { message: 'Hello, world!' } });

wait

Waits for an operation to complete.

Props

Name	Type	Required	Description
id	string	Yes	The operation's ID to wait for
timeoutInSeconds	number	No	Timeout in seconds (defaults to 360)

Returns

Promise<Operation<any, any>> - The completed operation object

Example

const operation = await client.operations.wait({ id: '123' });

updates

Provides real-time updates for operations through subscription methods:

• updates.subscribe() - Subscribe to operation updates
• updates.unsubscribe() - Unsubscribe from operation updates

Example

const listen = (operation: Operation<any, any>) => {
    console.log(operation);
};
client.operations.updates.subscribe(listen);
client.operations.updates.unsubscribe(listen);

Types

Operation

Represents an operation in the system with its status and associated data.

Name	Type	Description
id	string	Unique identifier for the operation
type	OperationType	The type of operation (e.g. 'generate', 'upscale')
status	'pending' | 'finished' | 'failed'	Current status of the operation
input	Input	Input data provided when creating the operation
output	Output | null	Result data (null while pending or failed)
metadata	any	Additional metadata associated with the operation
createdAt	Date	Timestamp when the operation was created
updatedAt	Date	Timestamp when the operation was last updated

Type Parameters

Parameter	Description
Input	The type of input data specific to the operation type
Output	The type of output data (extends OperationOutput) for results

Operation Input Types

Each operation type has its own specific input requirements:

describeProduct

Input for creating product descriptions:

Name	Type	Description
imageId	string	ID of the image to describe

generate

Input for image generation:

Name	Type	Description
aspectRatio	AspectRatio	Desired aspect ratio ('1:1', '9:7', etc.)
productImageId	string	ID of the product image to generate from
prompt	string	Text prompt for image generation
quality	'low' | 'high'	Quality of the generated image
seed	number	Seed for randomness
version	1 | 2	Version of inference model to use

imagine

Input for imagination operations:

Name	Type	Description
aspectRatio	AspectRatio	Desired aspect ratio
prompt	string	Text prompt for imagination
seed	number	Seed for randomness

proposePrompt

Input for prompt proposals:

Name	Type	Description
productDescription	string	Description of the product to propose a prompt for
imageId	string	ID of the image to propose a prompt for

replaceArea

Input for area replacement:

Name	Type	Description
imageId	string	ID of the image to modify
maskImageId	string	Mask image ID
prompt	string	Text prompt for the replacement
strength	number	Strength of the replacement

retouch

Input for area replacement:

Name	Type	Description
category	'bottom' | 'dress' | 'top'	Category of the area to replace
imageId	string	ID of the image to modify
productDescription	string	Description of the product to use for the replacement
productImageId	string	ID of the product image to use for the replacement
seed	number	Seed for randomness

select

Input for selection operations:

Name	Type	Description
imageId	string	ID of the image to select

upload

Input for upload operations:

Name	Type	Description
imageId	string	ID of the uploaded image

upscale

Input for upscaling:

Name	Type	Description
imageId	string	ID of the image to upscale

Operation Output Types

Operations can return three types of outputs:

OperationOutputImageSingle

Output containing a single image:

Name	Type	Description
kind	'image/single'	Kind identifier
imageId	string	ID of the generated/processed image

OperationOutputImageMultiple

Output containing multiple images:

Name	Type	Description
kind	'image/multiple'	Kind identifier
imageIds	string[]	Array of image IDs
previewImageId	string	ID of the preview image

OperationOutputText

Output containing text:

Name	Type	Description
kind	'text'	Kind identifier
text	string	The output text

Each operation type returns a specific output type:

• describeProduct: OperationOutputText
• generate: OperationOutputImageSingle
• imagine: OperationOutputImageMultiple
• proposePrompt: OperationOutputText
• replaceArea: OperationOutputImageSingle
• retouch: OperationOutputImageSingle
• select: OperationOutputImageSingle
• upload: OperationOutputImageSingle
• upscale: OperationOutputImageSingle