Automating Google's Nano Banana Pro with n8n: 5 Real-World Visual Automation Examples

Shubham SHARMA

Introduction to Nano Banana Pro [00:00:00]

The presenter introduces Nano Banana Pro, Google's new image generation model, highlighting its potential as a "visual brain" rather than just an image generator. The video's main objective is to demonstrate how to automate Nano Banana Pro with n8n, an automation tool, to create five real-world visual automation examples. The goal is to inspire viewers to build their own systems to save time, produce visual content on the fly, and even close podcast guests.

Automation Demonstration Overview [00:00:00]

The video begins with quick demonstrations of the five automation examples:

• Personalized Podcast Thumbnails: An overview of using a Google Sheet with LinkedIn URLs to generate personalized podcast invitation thumbnails.

  

  

  

• Image Annotations: A preview of automatically creating detailed diagrams and annotations on objects and software interfaces.

  

  

  

• E-commerce Visual Reports: A look at generating daily infographics summarizing e-commerce sales data from a Google Sheet.

  

Automating Nano Banana with n8n [00:00:57]

n8n is presented as a powerful automation tool that allows users to connect various "nodes" (applications or services) to create sophisticated workflows. The core idea is to link data sources (like Airtable or Google Sheets) with Nano Banana Pro via n8n to automate image creation based on predefined prompts and input data.

Case 1: Image Annotations [00:01:23]

This first example demonstrates automating the generation of detailed annotations and diagrams on images.

• Inspiration: The presenter was inspired by physics diagrams overlaying handwritten elements on photographs for concept explanation [00:01:38].

  

• Initial Manual Process:
  • An image of an F1 car was uploaded to Gemini [00:01:49].
  • Gemini was prompted to describe how the original annotated image was made, providing a prompt [00:01:51].

    

• Automated Workflow with n8n and Airtable: [00:02:29]
  • The automation starts by retrieving a list of images from Airtable that need annotations [00:02:36].

    

  • Each image is sent to Nano Banana Pro via n8n for processing [00:02:42].

    

• Demonstrated Results of Generated Images: [00:04:22]
  • A skateboard image was annotated with detailed component explanations, demonstrating precise labeling [00:05:18].

    

    

  • A splitboard image was annotated with its features, showcasing the ability to detail complex items [00:05:26].

    

    

  • A smartwatch image had its components detailed through annotations, even removing the original watch in one instance and adding new annotations [00:05:34].

    

    

  • A charger image was annotated with its parts, proving versatility across different object types [00:05:44].

    

    

  • A camera image was annotated with its parts [00:05:51].

    

    

  • The complex Interactive Broker and n8n interfaces were automatically documented with explanations, showing AI's understanding of intricate software layouts [00:05:56].

    

    

    

    

  • The system successfully identified and annotated elements without explicit instructions, demonstrating its contextual understanding and ability to produce clean, professional diagrams.

    

Hosting n8n: Options [00:03:08]

The video discusses the different methods for hosting n8n to ensure continuous automation.

• Local Installation: Installing n8n directly on a local computer [00:03:03].
  • Pros: Free as it's an open-source tool.
  • Cons: Automations stop as soon as the computer is turned off, making it unsuitable for always-on workflows [00:03:07].
• n8n Cloud Service: Using n8n's official hosted service [00:03:14].
  • Cons: Can be expensive, with limitations on workflow executions (e.g., €24 per month for limited executions) [00:03:19].

    

• Virtual Private Server (VPS) Hosting: Hosting n8n on a self-managed VPS [00:03:31].
  • Pros: Cost-effective (potentially four to five times less expensive), allows for continuous automation without local machine dependency, and provides greater control over the instance [00:03:44].

    

Case 2: YouTube Visual Cheat Sheets [00:06:32]

This case focuses on automating the creation of visual cheat sheets (summaries) from YouTube video transcripts.

• Concept: Automatically create a visual "cheat sheet" for each new YouTube video that centralizes key content and makes it visually actionable for learning [00:06:46].

  

  

  

• Workflow Overview: [00:07:01]
  • The workflow begins with a form that takes a YouTube video URL as input [00:07:15].
  • n8n retrieves the transcript of the specified video [00:07:20].

    

  • The raw transcript data (initially in a less readable format) is processed by a code node to make it more suitable for Nano Banana Pro [00:07:47].

    

  • Nano Banana Pro is then prompted to generate an image, rendering the transcript as a cheat sheet in French, highlighting important information [00:08:03].

    

• Demonstrated Result: A cheat sheet was successfully generated for a video on MCPs, visually outlining client-server architecture, concrete examples, and future concepts, demonstrating its effectiveness for visual learning [00:08:15].

  

  

Case 3: Real Estate Home Staging [00:08:28]

This example addresses the needs of real estate agents for automatic floor plan generation and virtual home staging.

• Problem: Real estate agents often rely on basic photos and manually generated descriptions. There's a demand for automated 2D floor plans and AI-reworked, home-staged photos to enhance property listings [00:08:44].

  

  

• Workflow Setup: [00:09:10]
  • Property advertisement information (e.g., "4-room house, 150 m²") and raw, un-staged photos are uploaded to Airtable [00:09:16].

    

  • The n8n automation retrieves this information and the photos [00:09:37].

    

  • Image Creation 1 (Floor Plan): Nano Banana Pro is prompted to create a 2D floor plan based on the provided photos and textual description. The more precise the description, the more accurate the floor plan will be [00:09:47].
  • Image Creation 2 (Virtual Home Staging): Each original photo is sent to Nano Banana Pro with a prompt instructing it to act as a "3D architectural designer" and perform "virtual home staging" in a specified style [00:10:04].

    

• Demonstrated Results: [00:10:37]
  • A fairly basic floor plan was generated from the initial raw photos, with potential for greater precision given more input data [00:10:48].

    

  • An original, unkempt room photo was transformed into a beautifully staged living space, demonstrating AI's ability to interpret and enhance existing structural elements like windows [00:10:58].

    

    

  • This process highlights the potential for real estate agents to quickly generate professional marketing visuals directly from property viewings by simply uploading photos and dictating details [00:11:10].

Case 4: E-commerce Visual Report [00:11:36]

This use case shows how to automate the creation of daily visual e-commerce sales reports.

• Goal: Automatically generate a daily infographic summarizing sales, popular items, daily turnover, and other key statistics in a visually appealing and readable format [00:00:41].

  

• Workflow: [00:11:51]
  • The automation can be triggered daily (e.g., every day at 8 AM) [00:00:46].
  • Sales data is retrieved from a Google Sheet [00:11:56].
  • The data is then converted into a format suitable for Nano Banana Pro (e.g., CSV) [00:12:03].
  • Nano Banana Pro is prompted to act as a "data analyst" and create a "daily sales recap" infographic [00:12:08].

    

    

  • The generated image is then inserted into a specific cell (e.g., "stat") in the Google Sheet, or could be sent via email or direct message platforms like Slack [00:12:34].
• Demonstrated Result: An infographic was generated showing "Daily Sales Recap - Oct 25, 2025" with total revenue, most active region, active category, and acquisition sources [00:12:54].

  

Case 5: Personalized Podcast Thumbnails [00:13:42]

This final example illustrates creating hyper-personalized podcast thumbnails to improve outreach for guest invitations.

• Challenge: Inviting high-demand podcast guests requires a unique and compelling approach to capture their attention and ensure they read the invitation message [00:13:45].

  

• Solution: Create a "wow effect" by sending a message that includes a custom thumbnail featuring the potential guest alongside the host, as if they've already been on the podcast [00:13:47].
• Workflow: [00:14:01]
  • A Google Sheet lists the LinkedIn URLs of desired podcast guests [00:14:10].

    

  • The Apify "LinkedIn profile scraper" is used to automatically retrieve the guest's name and profile picture from their LinkedIn profile [00:14:13].

    

  • Nano Banana Pro receives two images: a generic podcast template image (showing the host and an empty guest seat) and the guest's retrieved profile picture [00:14:48].

    

  • The prompt instructs Nano Banana Pro to combine these, creating a realistic screen split, placing the guest's face in the empty seat, and refining details like lighting and microphones for a natural look [00:14:56].

    

    

• Demonstrated Results: Shows impressive, realistic thumbnails of the presenter with several high-profile individuals, effectively simulating a podcast interview [00:15:49].

  

  

Technical Integration Details with n8n [00:16:57]

The presenter dives into the technical reasons behind using specific tools and methods for n8n integration, particularly regarding Nano Banana Pro.

• Why Kai Instead of Gemini Directly for Image Creation: [00:17:05]
  • The n8n "Generate Image" node currently uses an older version of Nano Banana, not the latest "Pro" version that offers superior results.
  • Kai acts as a token reseller and API aggregator, providing access to the latest version of Nano Banana (and other AI models) through a single, consolidated API [00:17:25].

    

• Integrating Kai's API with n8n via HTTP Request Node: [00:17:57]
  • Kai provides curl examples for its API calls, which can be easily imported into n8n's HTTP Request node [00:18:09].

    

  • Users then only need to dynamically replace specific parameters, primarily the "prompt," with data from preceding nodes in the n8n workflow [00:18:34].

    

• Asynchronous Task Management (Polling): [00:18:53]
  • Image generation by Nano Banana Pro is an asynchronous process, meaning it's not instantaneous.
  • n8n implements a polling mechanism: after initiating an image generation task, a loop node (e.g., "Attendre") periodically checks the status of the task (e.g., every 5 seconds) until it's reported as complete [00:18:56].

    

    

Conclusion [00:20:10]

The presenter reiterates the immense potential of automating Nano Banana Pro, encouraging viewers to explore and implement similar solutions in their own businesses or activities. The significantly improved quality of output from Nano Banana Pro, particularly its realistic generation of human faces, is highlighted as "insane" and production-ready, marking a significant advancement over previous AI image generators. Viewers are invited to share their own automation ideas in the comments and are directed to another video on Sora automation for further learning.