0

ACT 3 AI

Automated pipeline for generating Blender videos with motion capture and audio-driven lip-sync using distributed computing.

ACT 3 AI is a sophisticated video generation pipeline that automates the creation of Blender videos using YAML-based patterns derived from user-provided data. The system integrates advanced motion capture and lip-sync technologies within a distributed computing architecture.

Key Features

Automated Video Generation

  • Designed and implemented an automated pipeline to generate Blender videos using YAML-based patterns
  • Successfully integrated full-body motion capture for realistic character animations
  • Implemented audio-driven lip-sync for 3D characters within Blender

Distributed Computing Architecture

  • Built a distributed master-slave job execution system using Temporal
  • Leveraged on-premises PCs as compute nodes for scalable video rendering
  • Optimized rendering workflows for efficient resource utilization

Advanced Rendering Capabilities

  • Blender scripting using Python for automated scene generation
  • Integration with ComfyUI for enhanced AI-driven workflows
  • Video processing using ffmpeg for output optimization

Technical Highlights

  • Blender automation using Python scripting
  • Motion capture data integration for realistic character movement
  • Audio analysis and lip-sync generation
  • Distributed job scheduling with Temporal
  • High-performance rendering pipelines
  • Integration with Nvidia Omniverse and Unreal Engine

Built with

  • Blender - 3D creation suite
  • Python - Blender scripting and automation
  • C++ - Performance-critical components
  • Fast API - Modern Python web framework
  • Temporal - Distributed job orchestration
  • ComfyUI - AI workflow interface
  • FFmpeg - Video processing
  • Nvidia Omniverse - Real-time 3D collaboration
  • Unreal Engine - Real-time 3D creation tool