Hey, I have a JSON with multiple image URLs, followed by an Iterator to get each URL separately, then I’m using “HTTP Get a File” to get those images as jpg. I want to upload each image to one Claude message.
When I do that like in the attached screenshot, where I’m able to upload multiple images, my issue is: How can I define to first use image no 1, then image no 2, etc.?
My alternative solution attempts so far, each didn’t work:
I either want to merge 5-8 images into one and upload it to have Claude read the image. I don’t know how to do that.
Or I could use ChatGPT or an OpenAI Assistant - here’s the challenge that while I can add files consecutively to a Vector Store… jpgs are not supported and I would have to make them into pdfs for example.
I guess my question is: how can I have each image with its own name after getting them separately?
Coming back to this workflow after some time I’ve built a workaround for anyone having a similar challenge. I’m using the “Analyze Images (vision)” ChatGPT Module inside an Iterator-Aggregator to get my images described. After the Aggregator I’m posting the combined description of the images to Claude to continue working with it.
Maybe not the cleanest solution, because Claude never sees the image and only gets a description of it from ChatGPT, but it helps me progressing with the workflow for now.
It’s me once again Found the best solution (I think): I use an Iterator to go over each image URL, after that I use the HTTP Get A File module to get all images separately, then aggregate, and then target the image name and image “data” for each image. I know there are 5-10 images each time so I just set the Claude module up to attach 10 images to my prompt.
Not sure if this was even available when I first tried. But anyway, it’s exactly what I needed.
I’m having huge issues with this exact automation also… and can’t figure out what I’m doing wrong even after trying your solution. I’m positive its user error I feel like I’ve tried every single combination of settings from the Array aggregator through to GPT and/or Anthropic and can’t get it working.
Wondering if you could help.
This is the automation… currently want to pass the images through both GPT and Anthropic… but let’s start with just GPT.
This is the output from GPT. It appears that the input is not an array of images, just one image. And the output confirms that with ““Number of images recieved”” 1
I’ve built a workaround for anyone having a similar challenge. I’m using the “Analyze Images (vision)” ChatGPT Module inside an Iterator-Aggregator to get my images described. After the Aggregator I’m posting the combined description of the images to Claude to continue working with it.
I feel like I’ve tried every single combination of settings from the Array aggregator through to GPT and/or Anthropic and can’t get it working.
