Skip to main content

Welcome to Discomfort

Discomfort is a powerful ComfyUI extension that revolutionizes how you work with AI workflows by enabling programmatic execution with loops, conditionals, and persistent state management.

The Problem​

ComfyUI's native execution model follows a Directed Acyclic Graph (DAG) pattern, executing nodes once in topological order. While powerful for single-pass workflows, this model doesn't natively support:

  • 🔄 Iterative refinement workflows
  • ⚡ Conditional execution paths
  • 💾 State preservation across iterations
  • 🧩 Dynamic workflow composition
  • 🐍 Programmatic access and management

You spend 95% of your time building "Comfy spaghetti" instead of just creating. What if you could just write a script that tells ComfyUI exactly what to do?

The Solution​

Discomfort addresses these limitations by introducing an execution layer on top of ComfyUI's DAG model, using:

  • Pre-execution graph manipulation for dynamic workflow modification
  • Self-managed ComfyUI server for isolated runs
  • Simple data store that handles context throughout execution
  • Intelligent pass-by-value/reference system for different data types

✨ Key Features​

🐍 Loops & Conditionals​

for i in range(10):
    if i % 2 == 0:
        await discomfort.run(["empty_latent.json"])
    else:
        await discomfort.run(["img2img.json"])

💾 Persistent State Management​

with discomfort.Context() as context:
    context.save("model", my_model)  # Saved across runs
    await discomfort.run(["workflow.json"], context=context)
    result = context.load("output_image")

🧩 Workflow Stitching​

workflows = ["load_model.json", "process.json", "save.json"]
stitched = discomfort.Tools.stitch_workflows(workflows)
await discomfort.run([stitched["stitched_workflow"]])

🔌 Simple Integration​

Just add DiscomfortPort nodes to your existing ComfyUI workflows - no complex modifications needed!

🚀 Quick Start​

1. Install Discomfort​

cd ComfyUI/custom_nodes
git clone https://github.com/Distillery-Dev/Discomfort.git discomfort
cd discomfort
pip install -r requirements.txt

2. Add DiscomfortPorts to Your Workflow​

Open any ComfyUI workflow and add DiscomfortPort nodes:

  • INPUT ports: Connect output → workflow (inject data from Python)
  • OUTPUT ports: Connect workflow → input (extract data to Python)
  • Give each port a unique unique_id

3. Write Your Script​

import asyncio
from custom_nodes.discomfort.discomfort import Discomfort

async def main():
    discomfort = await Discomfort.create()
    
    with discomfort.Context() as context:
        # Set initial parameters
        context.save("prompt", "A beautiful landscape")
        context.save("seed", 12345)
        
        # Run workflow
        await discomfort.run(["my_workflow.json"], context=context)
        
        # Get results
        result_image = context.load("output_image")
        discomfort.Tools.save_comfy_image_to_disk(result_image, "result.png")
    
    await discomfort.shutdown()

asyncio.run(main())

📚 What's Next?​

🎯 Perfect For​

  • 🔄 Iterative refinement: Progressively improve outputs through multiple passes
  • 🎛️ Parameter sweeps: Test multiple configurations automatically
  • 📊 Batch processing: Process large datasets with custom logic
  • 🧪 A/B testing: Compare different approaches systematically
  • 🏗️ Complex pipelines: Build sophisticated multi-stage workflows

⚡ Current Status​

Discomfort is in alpha but fully operational. Core functionality including:

  • ✅ Discomfort Class - Complete programmatic API
  • ✅ Context Management - State persistence and data handling
  • ✅ Workflow Stitching - Merge multiple workflows seamlessly
  • ✅ ComfyUI Integration - Stable server management

Ready to transform your ComfyUI workflows? Let's get started!