Help with Integrating Meshy AI and OpenAI API in 8th Wall Backend Module

🧑‍💻 Technical Support
, ,
1

Hi 8th Wall Support Team:

I’m currently working on a WebAR project using 8th Wall’s Editor version. The goal is to allow users to upload an image, which will be sent to Meshy AI’s image-to-3D API for model generation, and then render the generated GLB model back in the scene. Additionally, I intend to use OpenAI’s image analysis capabilities for further processing of uploaded images.

Here are the details of my setup:

  • Meshy AI Integration: I’ve set up the backend module (index.ts) and successfully send base64 images to Meshy’s API. I am receiving a task_id in response, but I’m running into trouble with polling the task status and retrieving the final GLB file.
  • OpenAI Integration: I am also exploring using OpenAI’s image analysis capabilities (e.g., object detection, scene analysis) in the workflow. I need clarification on whether OpenAI’s image analysis API can be efficiently integrated within the backend of 8th Wall or if a separate service is necessary.

My specific questions are:

Meshy API Questions:

  1. Does 8th Wall’s backend environment support long-running tasks like polling (e.g., 30–60 seconds with intervals) for Meshy’s image-to-3D API?
  2. Can I download and locally save the generated GLB file to /static/models/ within the backend function? If not, what are the recommended workarounds?
  3. Is fetch() compatible with await res.arrayBuffer() for downloading binary data inside a backend module in 8th Wall?

OpenAI API Questions:

  1. Can I integrate OpenAI’s image analysis API directly in 8th Wall’s backend, or should I use an external server for processing?
  2. Are there any guidelines on using OpenAI’s image-to-text API for scene/object detection within an AR project (similar to what I’m trying to do with Meshy)?
  3. What are the best practices for handling large-scale image processing (e.g., multiple images at once) in an 8th Wall environment with OpenAI’s API?

Any guidance or code examples would be greatly appreciated. I’m trying to confirm whether this functionality can be handled entirely within 8th Wall’s backend modules or if I need to use an external service for certain tasks.

Thanks in advance for your help!

2

8th Wall experiences are at their core client-side webpages with three.js.

That said you can most likely do this as the GLTFModel component takes in a URL as well and does the heavy lifting for you regarding loading.

In general when working with an API it’s best to do your calls async and await the response to update your experience.

1 Like
3

Thanks so much for your reply.