POV Video Automation Workflow (n8n)

This workflow automates the creation of short gen AI videos from prompts stored in a Google Sheet. It generates an image using Hugging Face FLUX, converts the image into a short video using RunwayML Gen-3 Turbo, waits until the rendering is complete, downloads the generated video, and stores it locally.

The gen AI shorts are them assembled into a single video, with an automated narration created with the TTS (text-to-speech) module in Python added to the final video.

Workflow Summary

The workflow performs the following operations:

  1. Reads a prompt from a Google Sheet.
  2. Checks whether the corresponding image already exists locally.
  3. If the image does not exist, generates it using Hugging Face FLUX.
  4. Saves the generated image to disk.
  5. Converts the image into a Base64 string.
  6. Submits the image to RunwayML Gen-3 Turbo for image-to-video generation.
  7. Waits while Runway renders the video.
  8. Polls the Runway API until the rendering is complete.
  9. Downloads the generated video.
  10. Saves the final video locally.

Workflow Diagram

The snapshot below shows the complete workflow inside n8n.

n8n Workflow

Detailed Node Description

1. Click to Exec WF

Starts with a manual trigger, for more control. Starts the WF with a user input (step number) through an n8n form.

2. Reads AI sheet

Looks up the corresponding row in a Google Sheet using the supplied sequence number and retrieves the prompt to be used for image generation (this node can be manually configured for touch-ups).

3. Check If Img Exists

Attempts to read the expected image file from disk. If the image already exists, image generation is skipped. Otherwise, the workflow generates a new image.

4. Code in JS

Extracts the prompt from the Google Sheet and prepares it for the subsequent API calls.

5. Gen HF image

Uses the Hugging Face inference API (FLUX.1-schnell) to generate an image from the prompt.

6. Write Img to Disk

Stores the generated PNG image locally.

7. Extract from File

Reads the image and converts it into Base64 so it can be submitted to the RunwayML API.

8. Run Runway ML

Sends the Base64 image together with the prompt to RunwayML Gen-3 Turbo to generate a short video (user credits needed).

9. Wait For Video

Introduces a delay before checking whether the video generation has completed.

10. Get video URL

Queries the Runway ML task endpoint to obtain the current rendering status.

11. Check Status

Determines whether rendering has completed. If not, the workflow waits again and repeats the status check.

12. Get Video bin

Downloads the finished video once Runway ML reports that rendering has succeeded.

13. Write Video To Disk

Saves the generated MP4 video to the local disk.

A Gen AI video created with this WF

POV: Rogue Earth

n8n Workflow JSON

You can download the workflow here: POV_YT_n8n_WF-No_Secrets.json
If you use it, please credit me.

Important

For security reasons the RunwayML API key has been removed from the shared workflow.

Replace the placeholder

Bearer YOUR_RUNWAYML_API_KEY

with your own RunwayML API key before running the workflow.