How do you automate YouTube videos with n8n?

An n8n YouTube automation is a workflow that runs on a schedule: it picks a topic, has an AI model write the script, calls an image API for every scene and a voice API for the narration, sends it all to a rendering service, and uploads the finished video with the YouTube node. It works, and it gives you full control. It also means five paid services, your own Google Cloud project, and a pipeline you have to fix when one of them changes. ClipFlux is the same pipeline already built into one app, with a scene-by-scene review before anything posts. Here is how the n8n version is put together, what it costs, and when each makes sense.

Step 1: Trigger the workflow and pick a topic

A Schedule Trigger node starts the run, for example every day at 9:00. The next topic comes from a Google Sheet you fill by hand, or from an AI model node asked for a new idea in your niche that is not already in the sheet.

Step 2: Write the script

An AI model node (OpenAI, Anthropic or another) writes the narration and returns it as JSON: one object per scene, with the line to speak and a description of the picture. Asking for JSON is what lets the next nodes loop over the scenes.

Step 3: Make a picture for every scene

An HTTP Request node sends each scene's description to an image API. Most image APIs answer with a job id, so the workflow loops: wait, ask whether the picture is ready, and continue. A 10-minute video needs dozens of pictures, at a few cents each depending on the model. Keeping the same character in every picture is the hard part: each request knows nothing about the others unless you send a reference image every time.

Step 4: Make the voice

A text-to-speech API, ElevenLabs being the common pick, turns each line into an audio file, which the workflow stores (Google Drive, S3) so the renderer can fetch it. Captions need the timing of every word, which some voice APIs return and some do not.

Step 5: Render the video

The pictures, audio and captions go to a rendering API, such as JSON2Video, Creatomate or Shotstack, as one JSON description of the timeline. The workflow polls until the MP4 is ready and downloads it. Camera moves, transitions and caption styling are all written into that JSON by hand.

Step 6: Upload it to YouTube

The YouTube node uploads the file with its title, description and tags. It signs in through your own Google Cloud project with OAuth. Uploads have their own default allowance of 100 calls a day, and setting a custom thumbnail costs 50 of the 10,000 daily quota units (Google's developer docs, 26 October 2026).

What it costs

For one 10-minute video a week (about 40 minutes a month), from each service's own pricing page on 26 October 2026:

That comes to roughly 60 to 75 dollars a month in services before your time. ClipFlux's Pro plan is $79 a month for 2,800 credits, about 7 ten-minute videos, with posting included. In money the two are close. The real difference is the setup, the upkeep and what the video looks like.

Where n8n pipelines break

• The YouTube sign-in expires. While a Google Cloud app's consent screen is in "Testing", its refresh tokens expire after 7 days (Google's OAuth docs, 26 October 2026), so uploads stop every week until you publish the app or sign in again.
• A job never finishes. Image and render APIs time out or fail now and then; without retry logic one stuck scene stops the whole run, and a blind retry can pay for the same picture twice.
• Characters change from scene to scene. Unless every image request carries the same reference and description, the narrator looks different in every picture.
• Nobody looks before it posts. A fully automatic workflow uploads whatever the AI wrote, wrong dates and all, unless you build an approval step in.
• An API changes. When a service renames a field or retires a model, the node fails until you update it.

n8n or ClipFlux?

Choose n8n if you already run n8n, want a format no tool makes, need a service ClipFlux does not use, or enjoy building the pipeline itself. Choose ClipFlux if you want the channel running this week, with long videos, consistent characters and a look at every video before it goes live. How a ClipFlux Channel posts by itself is on the YouTube automation page; the bigger picture is in What is YouTube automation?

Sources

• n8n pricing
• YouTube Data API quota costs
• Google OAuth 2.0 (refresh token expiry)
• ElevenLabs pricing
• JSON2Video pricing

Read on 26 October 2026. ClipFlux is not affiliated with n8n or any service named here.

Questions