Some brief thoughts on Apple's proposed 3D <model> web element
Via @mrdoob on twitter; Apple are proposing a new <model> element for the web that would allow devs to embed interactive 3D models, but not control anything inside that element (so no composing scenes from multiple models, adding/removing based on taps, changing materials etc.)
Apple's underlying ambition seems to be owning 3D across their hardware and software ecosystem, as the 'black-box' approach gives Apple a great deal of control to guarantee a coherent UX and content format across phones, laptops and future HMDs.
It would also guarantee interoperability between immersive computing contexts via a locked-down API : you pin a <model> in the corner of your apartment with your phone, it requests the spatial volume it needs and negotiates placement with other XR apps you're running. Then, when you pop on your Apple glasses its still there in the state you left it, like any other object in the room.
However this comes at the considerable cost of making <model> 'other' to the web itself. The magic of web standards is how open and composable they are - I can load a live WebRTC video stream and use it to texture a 3d model chosen by a javascript machine-learning model, and follows a text element around the screen, updated by a realtime data API.
In comparison, <model> would be a simple, inscrutable black-box; a bit like pdf viewers in chrome/safari, but able to be embedded directly in a webpage.
领英推荐
I understand Apple's reasoning, and the actual spatial UX across devices will feel very cool when the full-stack all works smoothly together. However, I think its a net negative for the web, as creators are forced to adopt new asset pipelines and create inert content thats always locked and limited to 'what it is', without the wild, messy, hackable, experimental edge that defines the web for me.
My small hope is that this isn't mutually exclusive with Apple pursuing full, open WebXR support (allowing us to chaotically mash together GLTFs and all manner of web content)
The proposed element: https://github.com/WebKit/explainers/tree/main/model
The original thread: https://twitter.com/mrdoob/status/1432257785442484226
Lead Product Designer
3 年I love the idea of pinning a <model> to a scene and having it live there no matter when you go to that location. This is something we are playing around with at @Historik. Being new to AR, I'm assuming this is currently possible but still a bit clunky. My greatest concern is how will this affect the UX between different devices? How much technical debt will that create between devices if an Android user wants to experience the same <model> in the same scene?
Founder, CEO of FactorySpec
3 年Very interesting and watching this very closely cheers for the post Ben Ferns