Unleash Boundless
Creative Possibilities

Design and deploy ComfyUI workflows for live video using Stable Diffusion
and real-time inference.

TRY PIPELINES
Read Documentation
Screenshot of a ComfyUI workflow interface showing real-time video transformation. The left panel shows a webcam input of a woman smiling with a coffee cup, connected to a middle prompt node reading 'Make me a 3D cartoon bunny holding a carrot', leading to the right panel showing the output - a 3D animated white bunny character holding a carrot against a pink background. The title reads 'Building AI Video with ComfyUI and just your Webcam

Play With Video AI Magic

Transform your videos in real-time with our hosted Pipelines demo. No setup required!

Try Pipelines

Key Features

Discover Pipelines

If you’re in search of inspiration, look no further. Try out Pipelines in our playground experience, download the source JSON, and build your own.

Try Pipelines
Graphical representation of ComfyUI Workflow

Visual Workflow Design

Use ComfyUI workflows to chain AI models like Stable Diffusion and SAM2 to create unique video outputs with endless possibilities.

Video AI Community

Open source is in our DNA. We build in public and love to collaborate with our community. Become a member of our Discord to join the movement.

Join Discord

Meet ComfyStream

Process video streams in real-time, optimized for high-performance inference.

ComfyStream GitHub
Get started
INTRODUCTION

What is a Pipeline?

A Pipeline is a customizable tool for connecting ComfyUI's AI workflows with Livepeer's video processing compute network, enabling real-time video transformation.

  • Chain together multiple AI models
  • No need for complex infrastructure management
  • Process video in real-time
  • Deploy to production with only a few clicks
Graphical representation of Livepeer Pipelines Information Tile

FAQs

What makes this solution production-ready?

The integration with Livepeer's infrastructure provides production-grade capabilities including automatic scaling, global distribution, and high availability with SLAs.

What can I build with the Pipelines?

You can build custom AI workflows for live video using Stable Diffusion, real-time inference, LoRA fine-tuning, and countless other models available through ComfyUI workflows.

Can I see examples of what others have built?

You can explore Pipelines here and join our Discord to see what others are creating and adding to the community.

How can I connect with the community?

You can join the community through Discord, Twitter, and GitHub.

How does ComfyUI's architecture benefit video processing?

ComfyUI's directed acyclic graph (DAG) structure naturally aligns with sequential video processing, making it ideal for complex video workflows.

How do I integrate with ComfyUI?

ComfyUI integration is built into the platform and can be implemented directly through JSON.

What GPU options are available?

No need to worry about GPUs as the Livepeer Network handles all the processing and scale needed to run any pipeline. If you are interested in using specific hardware please reach out to our team.

Is there a limit on the number of pipelines I can create?

No, you can create as many pipelines as needed for your projects both public & private.