Natural Language to Video JSON Prompts with GPT and Gemini for Veo 3

This n8n workflow simplifies the process of converting natural language descriptions into structured JSON prompts suitable for video generation, specifically tailored for platforms like Veo 3. It leverages the power of Large Language Models (LLMs) from Azure OpenAI, OpenRouter, or Google Gemini to interpret user input and format it into a usable video prompt structure.

What it does

This workflow automates the following steps:

1. Listens for Chat Messages: It acts as a listener for incoming chat messages, serving as the trigger for the entire process.
2. Processes with an LLM Chain: The received chat message is fed into a LangChain "Basic LLM Chain" which orchestrates the interaction with a chosen Large Language Model.
3. Generates Structured Output: The LLM processes the natural language input and generates a structured output, likely a JSON object, representing video prompts.
4. Parses Output: A "Structured Output Parser" node then takes the LLM's output and ensures it conforms to a predefined JSON structure, making it ready for downstream applications.
5. Supports Multiple LLMs: The workflow is designed to be flexible, allowing users to choose between Azure OpenAI, OpenRouter, or Google Gemini as the underlying language model for generation.

Prerequisites/Requirements

To use this workflow, you will need:

• n8n Instance: A running instance of n8n.
• LangChain Integration: The LangChain integration enabled in your n8n instance.
• AI Service Credentials:
  • Azure OpenAI: An Azure OpenAI account with an API key and deployment details.
  • OpenRouter: An OpenRouter API key.
  • Google Gemini: A Google Gemini API key.
• Understanding of JSON Structure: Familiarity with the desired JSON format for your video prompts will be helpful for configuring the "Structured Output Parser."

Setup/Usage

1. Import the Workflow: Download the provided JSON and import it into your n8n instance.
2. Configure Credentials:
   • Locate the "Azure OpenAI Chat Model", "OpenRouter Chat Model", and "Google Gemini" nodes.
   • For the LLM you intend to use, configure the necessary API credentials (API Key, region, deployment name, etc.) within the node's settings.
   • Note: You only need to configure the LLM you plan to use. You can disable or remove the others if not needed.
3. Configure the "Basic LLM Chain":
   • Ensure the "Basic LLM Chain" node is connected to your chosen LLM (Azure OpenAI, OpenRouter, or Google Gemini).
   • You may need to adjust the prompt or model parameters within the "Basic LLM Chain" to guide the LLM in generating the desired video prompt structure.
4. Configure the "Structured Output Parser":
   • Define the expected JSON schema or structure that the LLM should output. This node will validate and parse the LLM's response into a clean JSON object.
5. Activate the Workflow: Once configured, activate the workflow. It will now listen for incoming chat messages.
6. Send a Chat Message: Interact with the "Chat Trigger" (e.g., via a connected chat service or by manually triggering it in n8n) with a natural language description of the video you want to create (e.g., "Generate a prompt for a video about a cat chasing a laser pointer in a living room, with upbeat music and quick cuts.").
7. Review Output: The workflow will process your message, and the "Structured Output Parser" will output the generated JSON video prompt.