Hey there, thank you so much - that's a really helpful and considered response, I appreciate it! Just to double-check, before I go down this route instead: would the right approach here be to use SceneReconstructionProvider to get the physical environment mesh, render that with OcclusionMaterial, and leave my visible entities as normal virtual content without EnvironmentBlendingComponent? My assumption is that this would give me the behaviour I was originally expecting: real walls, furniture, etc. would occlude entities when they're genuinely in front entities in front of real-world geometry would stay visible entities would continue to depth-sort normally against Mac Virtual Display, and they'd stay visible inside a system Immersive Environment when using .coexist, rather than being treated as part of the passthrough layer. Does that sound like the right approach for this use case? That would get me most of the way there, with one final part I'm unsure about: if I use the reconstructed physical mesh for occlusion, is there an API that lets my app detect how far the user has dialled in to system environments under .coexist? For context, if an entity is positioned behind the user's sofa, I'd want it occluded while the sofa is visible in passthrough, but visible again once they dial into Bora Bora and the sofa is no longer part of their perceived environment. I found onImmersionChange, but the docs are making me think this is exclusively for the app's immersion. If the system environment's immersion isn't exposed, is there any other way to avoid invisible reconstructed geometry continuing to occlude content? Thanks once again, I really appreciate it! 🙏