Code Review or Architectural Advice

[wonderingtribe]

🏷️ Discussion Type

Question

💬 Feature/Topic Area

https://github.com/AI-WonderLand1/psychic-octo-fishstick

Feed

Body

I'm looking for an efficient pipeline to stream assets (NPCs, environments, and scenes) between PlayCanvas and WebGL Studio. I'm using glTF-Transform for server-side optimization. How can I ensure that metadata (like NPC scripts or custom properties) and scene hierarchies remain consistent when moving between these two engines? Has anyone implemented a 'headless' glTF optimization service that serves both? here's my site its not fully done yet only been me and idk couple months, https://dreammakerhub.website/homepage , please anything would be nice I've never coded before thank you.

[NikhilJain-08]

Hi @wonderingtribe! 👋

That sounds like an awesome engineering project. You can definitely achieve this by building a centralized asset optimization pipeline using glTF-Transform.

Since you need to maintain consistency across both engines, here is the basic workflow framework you would use to make it happen:

1.Leverage glTF extras: Store your custom NPC scripts and hierarchy metadata inside the extras object of your glTF nodes, as standard optimization steps won't strip these out.

2.Build a Headless Service: Set up a lightweight Node.js server (using Express) running @gltf-transform/core and @gltf-transform/functions.

3.Run Optimization Passes: Have your server programmatically run passes like dedup() and prune() whenever an asset is requested or uploaded.

4.Stream to Runtimes: Serve the finalized, compressed GLB files directly to PlayCanvas and WebGL Studio, ensuring both client-side loaders are configured to read the extras field.

[asaddevx]

Hey @wonderingtribe,

This is a big task for someone who has never coded before, but it's doable with some planning.

The short answer

Yes, you can build a headless glTF optimization service. The key is using the extras field in glTF files. This field stores custom data like NPC scripts and scene hierarchies, and it survives optimization passes (like dedup and prune) that glTF-Transform runs.

The basic flow

1. Your server (Node.js + Express) runs glTF-Transform functions (dedup(), prune(), etc.) on uploaded assets
2. Your custom metadata stays safe inside the glTF extras object
3. Both PlayCanvas and WebGL Studio read that same extras field to rebuild your NPCs and scene structure

For a beginner

Start small:

• Get one cube with a custom property moving between engines first
• Add your NPC script metadata next
• Scale up to full scenes last

Your site at dreammakerhub.website shows real effort. Keep going, and good luck.