ChatGPT Image Upload

[SiteAltitude]

Are we able to automate ChatGPT with the image upload feature? My use case is describing images and I have a list of Image URLs and I'd love to upload those to a spreadsheet, have ChatGPT describe them and then add those descriptions back to the spreadsheet. Currently, the manual way to do it would be to visit each image URL, save the image, upload the image to ChatGPT and then copy the output.

 

[ArshilAhmad]

Hi @SiteAltitude,

Please allow me some time to discuss this with my team. I will get back to you with an update soon.

 

[Fagun Shah]

Are we able to automate ChatGPT with the image upload feature? My use case is describing images and I have a list of Image URLs and I'd love to upload those to a spreadsheet, have ChatGPT describe them and then add those descriptions back to the spreadsheet. Currently, the manual way to do it would be to visit each image URL, save the image, upload the image to ChatGPT and then copy the output.

This hasn't been added to OpenAI APIs yet. So not possible for now via Pabbly Connect. But you can try RPA tools like Robomotion to do this.

 

[ArshilAhmad]

As Fagun rightfully pointed out, the use case you are describing is not possible with ChatGPT currently. We do have Google Cloud Vision, but that won't completely suit your use case.

 

[Cameron Perrin]

Could we look into this again? ChatGPT has since released its vision model which can detect images. I set up this exact workfow in Zapier and it's working great but I'd love to switch to Pabbly.

 

[ArshilAhmad]

Could we look into this again? ChatGPT has since released its vision model which can detect images. I set up this exact workfow in Zapier and it's working great but I'd love to switch to Pabbly.

You can try using 'gpt 4 vision preview' AI Model to achieve your use case.

 

[Cameron Perrin]

You can try using 'gpt 4 vision preview' AI Model to achieve your use case.
View attachment 37370

I just ran a test twice and it seems to be looking at a random image instead of the link to the google drive image. Here are both runs from the same image (which is just a test image of Stitch from Lilo and Stitch), it's generating random titles every time that have nothing to do with the image.

 

[ArshilAhmad]

Please try mapping the preview link in ChatGPT action step and see if this gives you the desired result.

 

[Cameron Perrin]

Here is what it shows with the preview link. This is a photo of a desert

 

[ArshilAhmad]

Please allow me some time to discuss this with my team. We will get back to you with an update soon.

 

[ArshilAhmad]

I had a discussion with the technical team, and they informed me that the use case you are describing is currently not possible. We do not currently support obtaining image details through ChatGPT.

 

[Himesh]

Hello @Cameron Perrin you can download vision app from here it'll suit your use case

https://store.effortlessautomate.com/product/openai-vision

FYI I'm not from Pabbly connect team I'm just freelancer who help other Pabbly connect members.
Feel free to dm me if you need any help.