Livepeer-App-Pipelines Overview

This repository contains several example applications demonstrating different use cases for Livepeer's container pipeline infrastructure:

📁 batch/ - Batch Processing Examples

Contains two example applications for batch processing workloads using Livepeer's BYOC infrastructure:

• batch/create-a-tall-tale/ - Text generation application showcasing streaming SSE output of batch workloads with Livepeer BYOC integration
• batch/image-gen/ - Fast image generation application demonstrating 100s of millisecond response times using AI models

📁 byoc-stream/ - Real-time Video AI Processing

A comprehensive real-time video processing application that enables live AI processing on video frames using BYOC streaming. Includes:

• Gateway & Orchestrator - Livepeer infrastructure components
• Web Application - Main interface for stream management
• Multiple AI Workers examples:
  • frame-skipper/ - Frame skipping optimization for streaming
  • generate-video/ - Video generation from no input
  • passthrough/ - Basic passthrough processing pipeline
  • video-analysis/ - Real-time video analysis capabilities
  • webrtc-to-trickle/ - Bridge local webrtc enabled apps through Livepeer BYOC for scaling

📁 live-video-to-video/ - Live Video-to-Video Processing

A real-time video processing application focused specifically on video-to-video transformation using AI models through the Livepeer network.

• Complete streaming infrastructure with local testing interface

Setup Requirements

See the individual README files in each folder for specific setup instructions. Common requirements:

• Docker and Nvidia Container Toolkit installed
• Access to Nvidia GPU for AI processing
• (OPTIONAL) Ethereum address with deposit and reserve for payments (https://docs.livepeer.org/gateways/guides/fund-gateway)
  • use price of 0 if working with unfunded Gateway

Payment for compute is based on time the request takes to complete. For streaming, the payment is based on time the stream is live.