Text2Video Workflow

Cinna Text-to-Video Workflow

Generate high-quality videos directly from text using Cinna’s distributed AI infrastructure. This workflow extends ComfyUI functionality and is executed on decentralized GPUs coordinated through Solana.

---

Overview

The text-to-video workflow allows you to describe a scene in natural language and receive an AI-generated video as output. The process runs across compute providers in the Cinna network, ensuring scalability, transparency, and efficiency.

---

Supported Models

Workflow ID	Model
1	Hunyuan

---

Before You Begin

1. Request your Cinna API key

2. Configure the following in your .env file:

envCopy codeCINNA_API_KEY=your_api_key
CINNA_WORKFLOW_URL=https://sequencer-v2.cinna.ai

---

Example Request

tsCopy codeimport Cinna from 'cinna'

const cinna = new Cinna({
  apiKey: process.env['CINNA_API_KEY']
})

async function generateVideo() {
  const task = new Text2VideoTask({
    workflow_id: '1',
    prompt: 'a rabbit moving quickly in a beautiful winter scenery nature trees sunset tracking camera',
    timeout_seconds: 600
  })

  const response = await cinna.workflow.executeWorkflowAndWaitForResult(
    task,
    600000,
    15000
  )

  console.log('Generated video URL:', response.result)
}

---

Example Response

jsonCopy code{
  "task_id": "7599f1f038",
  "create_timestamp": 1733965543,
  "inference_latency": 0.608921,
  "upload_latency": 2.316333,
  "task_type": "txt2vid",
  "status": "finished",
  "result": "https://sequencer-v2.cinna.ai/files/generated-video.webp",
  "task_details": {
    "parameters": {
      "fps": 24,
      "height": 480,
      "length": 37,
      "prompt": "a rabbit moving quickly in a beautiful winter scenery nature trees sunset tracking camera",
      "quality": 80,
      "seed": 704883238463297,
      "steps": 30,
      "width": 848
    }
  },
  "timeout_seconds": 600,
  "workflow_id": "1"
}

---

Parameters

Type: Text2VideoTask

Property	Type	Required	Description
workflow_id	string	yes	ID of the video generation workflow
prompt	string	yes	Text description of the video
width	number	no	Width of the output video (default is 848)
height	number	no	Height of the output video (default is 480)
length	number	no	Length of the video in frames (default is 37)
steps	number	no	Inference steps (default is 30)
seed	number	no	Optional seed for reproducibility
fps	number	no	Frames per second (default is 24)
quality	number	no	Output quality (default is 80)
job_id_prefix	string	no	Optional identifier prefix
timeout_seconds	number	no	Maximum duration before the task is canceled
consumer_id	string	no	Optional consumer ID override
api_key	string	no	Optional API key override

---

Return Type

Type: WorkflowTaskResult

Property	Type	Description
task_id	string	Unique task identifier
status	enum	Current status of the task
result	any	Output URL or content
task_details	any	Input parameters and metadata
timeout_seconds	number	Maximum duration allowed
workflow_id	string	ID of the workflow used
create_timestamp	number	Task creation time
inference_latency	number	Duration of model inference
upload_latency	number	Time taken to upload the result

---

Authentication

Cinna uses a combined format for authentication: consumerId#apiKey. You can pass this information in two ways:

Option 1: During client initialization

tsCopy codeconst cinna = new Cinna({
  apiKey: process.env['CINNA_API_KEY'] // Format: consumerId#apiKey
})

Option 2: In the task itself

tsCopy codeconst task = new Text2VideoTask({
  consumer_id: 'yourConsumerId',
  api_key: 'yourApiKey',
  ...
})

---

Task Class Definition

tsCopy codeclass Text2VideoTask extends WorkflowTask {
  constructor(options: Text2VideoTaskOptions)

  get task_type(): WorkflowTaskType.Text2Video

  get task_details(): {
    parameters: {
      prompt: string
      width?: number
      height?: number
      length?: number
      steps?: number
      seed?: number
      fps?: number
      quality?: number
    }
  }
}

---

Workflow Task Type

tsCopy codeenum WorkflowTaskType {
  Text2Video = 'txt2vid'
}