AI Carousel Replication and Translation - n8n Workflow

Social Media Managers who localize content from global markets.
Content teams needing to quickly replicate high-performing visual assets.
Users seeking advanced examples of AI vision, translation, and image generation within an n8n workflow.
Agencies looking for robust n8n templates for content adaptation.

This advanced n8n workflow solves the problem of manually monitoring, translating, and recreating multi-image social media content (specifically 3-image TikTok carousels). By leveraging multiple AI services, this automation provides a significant reduction in labor and time. The n8n automation starts with a scheduled n8n trigger, monitors a TikTok RSS feed, and uses Google Gemini's Vision capabilities to analyze the exact text, layout, and composition of the original images. The text is translated via OpenAI, and a detailed prompt is generated to instruct Midjourney (or a similar service) to recreate the carousel set with the new, translated text perfectly embedded in the original layout. The final output is an organized Google Sheet and an email notification, making this a highly efficient n8n workflow for content scaling. This specific n8n template showcases complex integration patterns.

The process begins with the Daily TikTok Check n8n trigger, which runs on a schedule.

1. Monitor Feed: The n8n workflow fetches the specified TikTok account’s RSS feed using an HTTP Request n8n node.


2. Filter Posts: After parsing the RSS data, an If n8n node filters the results to ensure only posts identified as having exactly 3 images (carousels) are processed.


3. Process Images: The workflow splits the 3-image post into three separate items using the Split Out n8n node.


4. AI Vision Analysis: For each image, the Extract Text & Layout (Gemini Vision) n8n node downloads the image and uses Gemini-2.5-flash to extract critical visual data: text overlays, positional data, style, and overall image type. This is crucial for maintaining design consistency during recreation.


5. Translation: The extracted text is then translated into English using an OpenAI n8n node.


6. Prompt Engineering: The Create Midjourney Prompt Langchain Agent n8n node takes the detailed analysis from Gemini and the translated text, merging them to create an extremely precise Midjourney v6 prompt that demands exact layout and composition replication.


7. Image Generation & Wait: The n8n workflow sends the prompt to the Midjourney API. The Wait for Image Generation n8n node then halts the execution for up to 5 minutes, awaiting a webhook callback indicating the image is ready.


8. Aggregation: Once all three images are generated, the Combine 3 Images n8n node gathers the data (original/new URLs, translated text) for the complete carousel set.


9. Delivery & Notification: The final step uses the Google Sheets n8n node to save the complete record to a review dashboard and sends a detailed preview via the Gmail n8n node, completing the automated n8n workflow.

To install and run this powerful n8n workflow, follow these steps:

1. Import: Copy the provided JSON data and paste it into your n8n instance using the 'New' -> 'Import from JSON' function.


2. Credentials Setup: This n8n template requires several credentials. Ensure you have set up and configured connection credentials for:

1. Activation: Once credentials are set and configurations are complete, activate the n8n workflow.

Daily TikTok Check (Schedule Trigger n8n trigger): The starting point of this n8n workflow, set to run daily to monitor for new content.
Fetch TikTok RSS Feed (HTTP Request n8n node): Retrieves content from the specified TikTok account's RSS feed. Key configuration is the URL https://www.tiktok.com/@USERNAME/rss.
Filter: Only 3-Image Posts (If n8n node): Ensures continuity by only allowing items where imageCount (derived from previous parsing) exactly equals 3, focusing on carousel posts.
Extract Text & Layout (Google Gemini n8n node): Performs complex image analysis using the models/gemini-2.5-flash model. It extracts text, position, style, and visual elements, outputting structured JSON data.
Translate to English (OpenAI n8n node): Uses GPT-4 to accurately translate the extracted foreign language text, ensuring high-quality output before recreation.
Create Midjourney Prompt (Langchain Agent n8n node): A critical n8n node where AI prompt engineering occurs. It uses analysis and translated text to create a detailed prompt that guides the image generation service to match the original composition.
Generate New Image (Midjourney) (HTTP Request n8n node): Sends the carefully crafted prompt to the Midjourney service via a POST request, including the required 9:16 aspect ratio.
Wait for Image Generation (Wait n8n node): Pauses the n8n workflow execution until the external Midjourney service returns a result via a dedicated webhook callback.
Combine 3 Images (Aggregate n8n node): Reassembles the separate image data streams into a single, cohesive item for delivery.
Save to Review Sheet (Google Sheets n8n node): Writes the final comparison data (Original URL, New Image URL, Translated Text, Image Type) into a Google Sheet dashboard for easy review.