Multi-image composition in n8n

Step-by-step

1. Download the source images from Google Drive using the Google Drive node
2. Merge the downloaded files into a single list
3. Extract the Base64 data from each binary file
4. Aggregate the Base64 strings into an images array
5. Define the prompt in an Edit Fields node
6. Use a Code node to define the 'parts' array and stringify the request
7. Pass the 'parts' array into an HTTP Request node to call the Gemini API
8. Convert the API response to a file and upload it back to Google Drive

Options

• Upload images directly via a form instead of downloading from Google Drive

Watch out for

• The 'parts' array must contain both the text prompt and each image's Base64 data for the API request to function

Tips

• Use a Code node to stringify the request to ensure the API receives the data in the correct format
• If the AI doesn't understand the initial request, go back and revise the prompt for better results

Highlights

“it's really insane that it was able to take me looking at a certain angle and put me looking down at a different angle... and incorporating me into the scene flawlessly”