AI Shorts Factory — Automated Motivational Video Generator

Overview

AI Shorts Factory is an end-to-end n8n workflow that produces short-form motivational videos with almost no manual editing. It pulls stories from Reddit, rewrites them into a video script, generates matching visuals and narration, and assembles a finished vertical video that can be uploaded straight to YouTube.

Making short-form content by hand means writing a script, sourcing images, recording voice-over, captioning and editing for every clip. This project replaces that loop with a single, configurable automation.

How it works

• An n8n form collects the run settings (subreddit, LLM provider, TTS engine, voice, music).
• Stories are fetched from Reddit and logged in Google Sheets so each one is tracked through the pipeline.
• An LLM rewrites the story and an information-extractor step breaks it into individual scenes with image prompts.
• Each scene gets an AI image from Flux.1-Schnell via Together AI and narration from Kokoro or Chatterbox TTS (with optional voice cloning), all running on a self-hosted Docker media server.
• Scenes are captioned, merged with background music, polled until rendering completes, then optionally uploaded to YouTube via the Google API.

Engineering notes

The workflow is large (around 95 nodes) and provider-agnostic: it ships with interchangeable chat-model nodes for OpenAI, Anthropic Claude, Google Gemini, Groq, DeepSeek, Mistral, xAI Grok, OpenRouter, Azure OpenAI, AWS Bedrock and Ollama. Switch/If routing, wait-and-poll loops, form-based error reporting and cleanup of temporary files keep long rendering jobs reliable.