ComfyUI Has a Free API — Build Complex AI Image Workflows Visually

#ai #machinelearning #opensource #python

ComfyUI: Node-Based Stable Diffusion Interface

ComfyUI is the most powerful GUI and backend for Stable Diffusion. Unlike simple text-to-image tools, ComfyUI lets you build complex workflows using a visual node editor — chain models, control nets, upscalers, and custom logic.

How ComfyUI Works

Every workflow is a graph of nodes. Each node does one thing: load a model, encode text, sample, decode, save. Connect them visually and you have a complete image generation pipeline.

The Free API

ComfyUI exposes a WebSocket + REST API:

import json
import urllib.request
import websocket

SERVER = "127.0.0.1:8188"

# Queue a workflow
def queue_prompt(workflow):
    data = json.dumps({"prompt": workflow}).encode("utf-8")
    req = urllib.request.Request(
        f"http://{SERVER}/prompt",
        data=data,
        headers={"Content-Type": "application/json"}
    )
    return json.loads(urllib.request.urlopen(req).read())

# Get generation history
def get_history(prompt_id):
    resp = urllib.request.urlopen(f"http://{SERVER}/history/{prompt_id}")
    return json.loads(resp.read())

# Download generated image
def get_image(filename, subfolder, folder_type):
    params = f"?filename={filename}&subfolder={subfolder}&type={folder_type}"
    resp = urllib.request.urlopen(f"http://{SERVER}/view{params}")
    return resp.read()

# REST API endpoints
curl http://localhost:8188/system_stats
curl http://localhost:8188/object_info  # All available nodes
curl http://localhost:8188/queue        # Current queue status
curl -X POST http://localhost:8188/interrupt  # Cancel current gen

Key Features

Node-based workflow — visual programming for AI
ControlNet — guide generation with poses, edges, depth maps
LoRA/LyCORIS — fine-tuned style models
Upscaling — Real-ESRGAN, Ultimate SD Upscale
Inpainting — selective image editing
Video — AnimateDiff, SVD workflows
Custom nodes — 1000+ community extensions

Real-World Use Case

A game studio needed to generate 500 character portraits with consistent style. ComfyUI workflow: SDXL base -> ControlNet (pose) -> LoRA (art style) -> face fix -> upscale. Run via API in batch mode. 500 images in 4 hours, consistent quality.

Quick Start

git clone https://github.com/comfyanonymous/ComfyUI.git
cd ComfyUI
pip install -r requirements.txt
python main.py --listen 0.0.0.0
# Open http://localhost:8188

Resources

Need automated AI data collection? Check out my scraping actors on Apify or email spinov001@gmail.com for custom AI automation.

DEV Community