Beginner’s Guide: NanoBanana in n8n - Create & Edit Images (No-Code Tutorial)

Key Concepts

• NanoBanana: Google's image generation and editing model.
• Naden: A no-code platform for building AI workflows.
• Text-to-Image Generation: Creating images from textual descriptions.
• Image Editing: Modifying existing images based on textual prompts.
• Open Router: A platform for accessing various AI models through a unified API.
• File.ai: A third-party platform that utilizes NanoBanana for image editing.
• HTTP Request Node: A Naden node for making HTTP requests to external APIs.
• Base64 Encoding: Representing binary data (like images) as a text string.
• Prompt Engineering: Crafting effective prompts to guide AI models.
• UGC (User Generated Content): Content created by users, often used in marketing.
• VPS (Virtual Private Server): A virtualized server that provides dedicated resources.

Generating Images from Scratch with NanoBanana in Naden

• Process:
  1. Trigger: A Naden form is used to collect a text description of the desired image.
  2. Prompt Generation: An OpenAI node (GPT-4o is used in the example) transforms the short user description into a detailed, hyperrealistic semantic prompt suitable for an image generation model. The prompt is designed to be understood by AI image generators.
     • Example prompt instruction: "Your task is as an advanced AI to specialize transforming a short provided description into hyperrealistic semantic first image prompt."
  3. Image Creation: An HTTP Request node sends a POST request to Open Router, specifying the NanoBanana model (2.5 flash image preview by Google Gemini) and the generated prompt.
     • The model name used is "2.5 flash image preview" which is by Google Gemini.
     • The modality is "image and text".
  4. Data Extraction: A function node extracts the image URL and the Base64 encoded image data from the HTTP response.
  5. Base64 Decoding: A "Convert B64 to file" node converts the Base64 string into an actual image file that can be downloaded.
• Example: The user inputs "an image of a 22-year-old blonde model holding a red can of sliced soda." The workflow generates an image of a model holding a can of Slice soda.
• Technical Details:
  • Open Router is used as an intermediary to access the NanoBanana model.
  • The HTTP Request node uses the POST method to send the prompt to the Open Router API endpoint.
  • Base64 encoding is used to represent the image data as a text string.

Editing Existing Images with NanoBanana in Naden

• Process:
  1. Trigger: A Naden form collects both an image file and a text description.
  2. Image Upload: The uploaded image is stored in Google Drive.
  3. Image Description: An OpenAI node analyzes the uploaded image and extracts details such as brand names, hex codes, font styles, and visual descriptions.
     • The OpenAI node is instructed to "analyze the image and basically give us the brand name, the hex code, the name, the font style, the visual description".
  4. Prompt Generation: Another OpenAI node generates a new prompt that combines the user's text description with the AI-generated description of the existing image. This prompt guides the image editing process.
     • The system prompt instructs the AI to act as a "UGC (user generated content) image prompt builder".
  5. Image Editing: An HTTP Request node sends a POST request to the File.ai API, providing the generated prompt and the URL of the uploaded image. File.ai uses NanoBanana to edit the image according to the prompt.
  6. Image Retrieval: The workflow waits for File.ai to process the image and then retrieves the URL of the edited image.
  7. Image Download: The edited image can then be downloaded.
• Example: The user uploads an image of a can of Slice soda and provides the description "a 22-year-old blonde model holds a can of sliced soda." The workflow generates an image of a blonde model holding the same can of Slice soda, accurately replicating the can's details.
• Technical Details:
  • File.ai is used as a third-party platform to access NanoBanana's image editing capabilities.
  • The HTTP Request node sends the prompt and image URL to the File.ai API.
  • The workflow includes a delay to allow File.ai time to process the image.
• Key Argument: NanoBanana excels at capturing and replicating existing details of uploaded images, making it particularly useful for creating marketing ads.

Hosting Naden Workflows on Hostinger

• Recommendation: Hostinger is recommended as a cost-effective solution for hosting Naden workflows on a virtual private server (VPS).
• Benefits:
  • Unlimited workflows.
  • Full control and privacy.
  • Cheaper than Naden's cloud version.
• Steps:
  1. Sign up for a Hostinger account.
  2. Choose a VPS plan (KVM2 is recommended).
  3. Select the 24-month plan for the best discount.
  4. In the application settings, select Naden as the top application.
  5. Enter the coupon code "AI workshop" for an additional 10% discount.
  6. Complete the billing and payment information.
  7. Install the software.
  8. Access Naden through the Hostinger dashboard.
• Technical Details:
  • Hostinger provides a VPS environment for running Naden.
  • The "AI workshop" coupon code provides a 10% discount.

Conclusion

The video demonstrates two powerful applications of Google's NanoBanana within the Naden no-code platform: generating images from scratch and editing existing images. The presenter highlights NanoBanana's ability to accurately replicate details from existing images, making it a valuable tool for marketing and advertising. The video also provides a practical guide to hosting Naden workflows on Hostinger, offering a cost-effective alternative to Naden's cloud-based solution. The key takeaway is that NanoBanana, combined with Naden's no-code interface, empowers users to create and manipulate images with ease, opening up new possibilities for AI-driven content creation.