🤖 Your AI Video Factory, Powered by n8n

Built with n8n
, , ,
1

Hey folks,

I’ve been following the rise of AI video generation closely, and while the tech is mind-blowing, I noticed one major issue — creating videos manually, one after another, is slow and repetitive. I wanted a way to turn a single creative spark into a full batch of videos, ready to go.

That’s why I built something new in n8n: the AI Director :clapper_board:

It’s essentially a mini production studio inside a workflow — designed to take one idea and automatically expand it into a collection of short-form videos with their own matching audio tracks.

How it Works

The workflow uses a two-agent setup with Gemini, then branches out for parallel generation:

:performing_arts: Creative Writer Agent – Starting from a Form Trigger where you input your concept (genre, tone, mood, etc.), this agent builds a full storyline with screenplay details, character descriptions, and camera directions.

:film_projector: Scene Breakdown Agent – The draft is then broken into multiple distinct scenes, each formatted as a prompt for short clips (perfect for 15–30s Reels or Shorts).

:high_voltage: Parallel Generation – Each scene is processed in two tracks at once:

  • One goes to Google Gemini to generate the video.

  • The other goes to ElevenLabs to produce an original background audio layer.

At the end, you get a batch of video + audio pairs, all born from that one initial concept.

Example Outputs

Here’s a taste of what it can make:

Key Highlights

:sparkles: Batch Production – Generate a whole content library from one idea.
:robot: AI Production Crew – From storyboarding to shot planning, it’s all automated.
:headphone: Custom Audio – Each clip gets a synchronized soundtrack.
:artist_palette: Flexible Controls – Adjust genre, tone, or animation style right at the input step.

This is especially useful for creators, marketers, or anyone in the n8n community who wants to experiment with large-scale creative automation.

I’ve bundled the workflow with a full guide and published it on Gumroad here:
:right_arrow: Video Stack Generator

:warning: Note on Costs: This does rely on paid APIs. The Gemini video API has usage fees, and ElevenLabs works best with a paid plan for batch runs.

Would love to hear your thoughts on this approach and if you have ideas for improvements. Also open to collaboration, workflow requests, or just connecting:
:envelope_with_arrow: Email | :link: LinkedIn

Thanks for checking it out!

1 Like
2

That’s a cool setup! I’ve been experimenting with n8n for automating some workflows and it’s pretty flexible once you figure it out. Curious what you’re using for the actual video generation part – ElevenLabs for voice? Runway for visuals?

One thing I struggled with was getting scripts that didn’t sound super robotic for voiceovers. Someone in another forum mentioned https://cleverhumanizer.ai/ai-writer and I gave it a shot. It’s basically an AI writer but focuses on making the output sound more conversational, which helps for video scripts.

I wouldn’t say it’s perfect – sometimes you still need to tweak phrases – but it’s definitely better than raw ChatGPT output for narration. The main benefit I found was it writes in a more natural speaking style rather than that formal AI tone. Could probably hook it into your n8n workflow for the script generation step if you’re looking to automate that part too.

How’s your current pipeline handling script quality? That’s been my biggest bottleneck tbh.