Using API to Call ComfyUI

1. Submitting Drawing Tasks

To submit a drawing task, you need to use the following POST request:

POST /prompt

Request parameters:

• client_id string A task ID generated by the client to identify the task initiator
• prompt json JSON data containing drawing parameters

Example

{
  "client_id": "unique_client_id",
  "prompt": {
    "width": 768,
    "height": 512,
    "text": "A beautiful landscape"
  }
}

2. Using WebSocket to Receive Task Status

After submitting the task, you can receive real-time updates via WebSocket. Connect to the following address:

ws://<your_server>:<port>/ws?client_id=unique_client_id

Once connected, you will receive information about the task execution status, progress, etc.

Data Format

• Text data: Used to notify task changes, current execution steps, and progress.
• Binary data: Used to transmit generated image previews.

Python Example Code

Below is an example code using Python and WebSocket client library to submit a drawing request and receive results:

import websocket
import json
import uuid

server_address = "127.0.0.1:8188"
client_id = str(uuid.uuid4())

def queue_prompt(prompt):
    p = {"client_id": client_id, "prompt": prompt}
    data = json.dumps(p).encode('utf-8')
    req = urllib.request.Request(f"http://{server_address}/prompt", data=data)
    response = urllib.request.urlopen(req)
    return json.loads(response.read())

def on_message(ws, message):
    print(f"Received: {message}")

ws = websocket.WebSocketApp(f"ws://{server_address}/ws?client_id={client_id}",
                            on_message=on_message)
ws.run_forever()

# Submit drawing request
prompt_data = {
    "width": 768,
    "height": 512,
    "text": "A beautiful landscape"
}
queue_prompt(prompt_data)