I’m looking a building a workflow to process Nano Banana image processing prompts via an HTTP requests. This will need to reference 4 different uploaded images, that will be referenced in the prompt. I cannot think of a way to upload the images for processing (via a base64 conversion) whilst being able to reference them by object name. Any ideas please?
What is the error message (if any)?
Please share your workflow
(Select the nodes on your canvas and use the keyboard shortcuts CMD+C/CTRL+C and CMD+V/CTRL+V to copy and paste the workflow.)
Before going too far with object names, I’d first check whether you really need the prompt to refer to the images by name. If you do, I’d still handle that in n8n metadata, not inside the image payload itself.
If you need named references, I’d map each uploaded image to a fixed role like image1 = product , image2 = background , image3 = style , image4 = logo , then send all 4 images in one Gemini request and describe them in the text prompt using that same order. Gemini supports multiple image inputs in one request, either as inline base64 or uploaded files. ()
If you do not actually need named references, the simpler solution is to send the 4 images as separate parts in one request and just say “use the first image as the subject, the second as the reference background,” and so on. That is usually cleaner than trying to make the API understand custom object names for each base64 blob. For n8n, I’d use inline base64 for small files, or the Files API if the images are larger or reused often. ()