Generate 9:16 Product Video Ads with OpenAI, Groq & JSON2Video

Generate 9:16 Product Video Ads with OpenAI, Groq & JSON2Video (Fully Automated in n8n)

Turn a product description and image into a complete vertical 9:16 AI product ad—complete with cinematic scene visuals, OpenAI voiceover audio, and automatic subtitles. This n8n workflow uses Groq, OpenAI, Google Drive, and JSON2Video to render the final video end-to-end.

What this workflow does

Designed for Form Trigger submissions, the workflow:

• Receives a product description and product image from an n8n Form Trigger.
• Uploads the submitted image to Google Drive and sends the description to Groq to generate three cinematic scene prompts.
• Downloads a reference image from Google Drive and uses OpenAI image editing (gpt-image-1-mini) to create three scene images, uploading each back to Google Drive.
• Generates a single-paragraph voiceover script (via Groq), then creates audio voiceover using OpenAI and uploads it to Google Drive.
• Compiles a JSON2Video project payload that assembles the three images into timed scenes with audio and subtitle settings.
• Sends the payload to JSON2Video, checks render status, downloads the final video, and returns it as the form completion response.

Use cases

• SaaS teams producing vertical product ads for listings, landing pages, and social campaigns.
• Automation engineers building an AI ad generation pipeline that outputs ready-to-publish 9:16 video creatives.
• E-commerce operators streamlining creative production from product data → ad video without manual editing.

Technical details

• Groq: Two Groq chat model steps (openai/gpt-oss-120b and openai/gpt-oss-20b) for scene prompts + voiceover script.
• OpenAI: gpt-image-1-mini for image generation and OpenAI audio generation for voiceover (or via n8n AI Gateway).
• JSON2Video: Renders the 9:16 video using a constructed JSON project payload with subtitles.
• Google Drive: Stores and retrieves the submitted image, generated scene images, audio, and the final video.
• n8n nodes referenced in the workflow: set, code, form, wait, merge, sticky note.