Project Overview:
Creating consistent short-form video reels manually requires hours of scriptwriting, scene planning, image generation, video rendering, voiceover synthesis, and editing. I engineered "AI Auto-Reel Factory"—a fully autonomous end-to-end video production pipeline in self-hosted n8n powered by multi-agent Qwen LLMs, Python orchestration, Image-to-Video AI, TTS voiceover generation, and automated FFmpeg rendering.
System Architecture & Multi-Agent Pipeline:

Complete "AI Auto-Reel Factory" n8n architecture featuring 4 specialized Qwen AI Agents with Think tools and Structured Output Parsers, Python scene batching, Image-to-Video generation, TTS audio synthesis, FFmpeg muxing, and Google Drive/Sheets sync.
How the Autonomous Reel Production Pipeline Works:
Script History & Deduplication: Reads previous scripts from Google Sheets (Get row(s) in sheet) and aggregates context so every generated reel is unique.
Multi-Agent Creative Suite (Qwen + Think + JSON Schema): Four chained AI Agents sequentially generate the viral script, break it into timed scenes/beats, craft detailed visual prompts, and write motion prompts—logging every stage into Google Sheets.
Scene-by-Scene Media & Voice Synthesis: A Python node prepares scene payloads for Loop Over Items, triggering API image generation, Image-to-Video synthesis, and automated TTS voiceover creation.
Automated Audio-Video Muxing & Final Assembly: Shell commands (Execute Command / Mux) synchronize voiceovers with generated video clips, stitch all scenes into a final 44MB reel, and upload assets directly to Google Drive.
Live Execution & Verified Production Run:

Verified production execution (ID#18602) completing full scriptwriting, multi-scene video synthesis, TTS voiceover muxing, and Google Drive upload in 10m 15s with green checkmarks across all active nodes.
Key Results & Technical Highlights:
Zero-touch short-form video factory transforming a single trigger into a fully voiced, edited, and rendered reel.
Strict Structured Output Parsers across all 4 Qwen AI Agents eliminate JSON formatting failures during multi-scene batching.
Self-hosted Docker n8n + Python + CLI video muxing architecture with zero recurring SaaS rendering fees.
0 likes · 0 views · Nov 16, 2026
Built an autonomous AI Auto-Reel Factory in n8n using multi-agent Qwen LLMs, Python, Image-to-Video AI, TTS, and FFmpeg muxing.