Passthru Camera -> off-device image processing (laptop server) -> send back to headset

I have a Quest 3 Unity prototype that captures passthrough frames, sends them to a laptop for processing, and returns full-frame processed images to the headset. That basic loop works, but the result is still clearly not equivalent to native passthrough.

My main question is: do public Meta APIs allow arbitrary processed frames to be injected into the true native passthrough compositor, or is the best possible public path a custom Unity-rendered approximation?

I’m especially trying to understand whether combining:

• passthrough RGB
• camera pose / intrinsics / extrinsics
• scene depth / raycasts

can ever produce a user-facing result that is effectively equivalent to normal passthrough, or whether that will always remain a custom approximation outside Meta’s internal native pipeline.

If native replacement is not publicly supported, a clear confirmation of that would be very helpful.

I’d also appreciate pointers to any existing libraries, sample projects, research repos, or community implementations that attempt something similar, especially if they go beyond a simple 2D panel/debug view and try to approximate a true full-view processed passthrough experience. Thanks!