Hello OpenAI Community,
I’ve encountered challenges sending DALL·E images via my Custom GPT action. It is understandable that Open AI is not allowing embedding DALL.E image URL into external website due to privacy concern, I’m exploring alternates to incorporate DALL·E images .
Here’s what I’ve attempted so far:
1. POST Request with Form-Data Multi-Part Binary File: My goal was to instruct GPT to convert the image into a binary format for transfer. However, despite clear instructions, the code interpreter seems to struggle or fails to accomplish this task. It appears that OpenAI’s custom actions might not fully support sending binary Image as Form-Data Multi-Part.
2. POST Request with Image Conversion and Chunking: This approach involved asking GPT to:
- Resize the image to a smaller dimension (e.g., 320px).
- Split the image into chunks, each containing 400 characters, formatted as a string array.
- Send this string array as one Post request.
Unfortunately, this method also led to the code analyzer getting stuck.
Has anyone successfully managed to send DALL·E images as part of a custom action in their GPT ? Any insights or alternative methods would be greatly appreciated.
For additional context, I’m working on a project called Presenty GPT (link: chat.openai.com/g/g-IumLsKslz-presenty). It’s designed to take user topics and, through custom actions, communicates with my website www.presenty.app to generate dynamic web-based presentations. Any advice to improve this integration would be incredibly valuable.
Looking forward to your suggestions and insights
Thank you.