Goal: I want to combine my jewelry products with different concept models using AI, and save the newly generated images into separate folders.
I upload my jewelry products into a Google Drive folder (Folder A - Image1).
I need to create a new folder in my Google Drive account (Folder name: Image1 - Product image 1-2-3…).
I already have folders in my Google Drive containing model images. Each folder contains different variations of images (Folder Model1 - Model image 1-2-3-4… / Folder Model2 - Model image X-Y-Z…).
I want the AI (Gemini) to randomly select the first jewelry product image and one of the model images from the first folder, and then combine them (Gemini: Product image1 & Model image1 - Product image1 & Model image X …).
I want the images generated in the 4th step to be uploaded into the folder created in Google Drive with the name “Product image1”.
Afterwards, I want the system to perform the same process for Product image2, pairing it with models in different variations.
Note: If I can get support for the workflow creation, I can fill in the node configurations and prompt details within the flow. Thank you everyone for your support.
What is the error message (if any)?
@Trilogy_Elektronik for the combine step ur gonna want Gemini 2.5 Flash Image (the “nano banana” model — it accepts multiple inline images + a prompt and outputs a composite). skeleton flow:
Google Drive — list files in folder A (jewelry products)
Loop Over Items — runs per product
Google Drive — create new folder named after the product
Google Drive — list files in a model folder, then a small Code node to pick one at random
HTTP Request to Gemini 2.5 Flash Image with both images as base64 + ur composition prompt
Google Drive — upload the returned base64 image into the folder from step 3
quick Q before I sketch any node JSON — are u on n8n cloud or self-hosted (and what version)? also are u set on Gemini specifically or open to other multi-image models? answers change the auth setup.
You already got a great high-level outline from @achamm, I’ll just add a few very practical notes that should make Gemini easier to plug in.
If you’re using Gemini, I’d recommend sticking to one of these models:
gemini-3.1-flash-lite for fast and cheap bulk generation
gemini-3.5-flash if you want the best quality and can afford a bit more cost
In n8n, instead of an HTTP Request node, use the built-in Google Gemini node:
Set the operation to Image → Edit Image so you can send both the product photo and the model photo, plus your prompt.
Create a Google Gemini (PaLM) API credential using an API key from Google AI Studio, then select that credential in the node.
Choose the model (gemini-3.1-flash-lite or gemini-3.5-flash) in the node settings, map your two input images, and you’re good to go.
The Gemini node will give you the result as a binary image, which you can pass directly into a Google Drive node to save it into the right product folder.
Thank you very much for your help. The automation is working now.
Next, I need to set up resolution 2K, ratio 1:1, temperature and top-p. I can’t find these parameters in the Gemini module. Do you know how to do this?
Best regards
Thanks a lot for marking my reply as the solution – really appreciate it, that means a lot
About your follow‑up question: I’d be very happy to help and walk through it in more detail, but to keep things clean and easier to find for others later, it would be best to open a new topic with the specifics of that issue. I’m online now and keeping an eye on the forum, so if you create a new, I can jump in and help you there right away.
