There are imperfect ways to test this out as a library in a 2D WebGL/WebGPU render on-screen, or in WebXR, but:
⢠In the former case a device like Meta Quest or Apple Vision Pro won't see the content in stereo,
⢠In the latter case, there's a requirement to vend the head- and eye-pose to the JavaScript, impacting privacy
⢠And in both cases, there's a significant indirection where it's necessary either to re-build CSS in WebGL, or to lean on other new proposals like HTML-in-Canvas. I do think that could be polyfilled for a non-spatial device/context, but believe it's the wrong way to proceed conceptually.