Introduce model to whatwg!

[zachernuk]

I've filed an issue here:
whatwg/html#10901
to begin the process of describing Model with the goal of landing the barest skeleton into the HTML spec. @marcoscaceres Has suggested we still discuss and deliberate here as 3D experts, but aim to put in changes there once we have consensus.

Perhaps the first question is how to indicate "implementor interest" - and for folks with a whatwg person internally, to discuss what other conversations might be necessary there too.

/agenda to check in with what folks need to do to register intent at whatever level necessary to proceed

[zachernuk]

This is on the docket for the WHATNOT Telecon on Thu, 30 January, 9am PT.
The scope is just to raise the idea of <model> itself, rather than adding any extra detail - but the goal is to get the element landed in the HTML spec!

@LaszloGombos @cabanier (and @alcooper91 etc if you're in a position to make any comment) - If you want to add anything to the issue or participate in the conversation, that would be great!

[cabanier]

Who can call into this meeting? Do I need to join WhatWG in order to comment?

[zachernuk]

@cabanier you should be able to join the call if you ask in the issue to be added - and commenting is open to all on the issue, there are no membership requirements 🥳

[alcooper91]

Thanks for this @zachernuk. I'm holding some more internal conversations at the moment and while I admit to not being fully aware of the WHATWG process and how long it takes, I do feel that there is still some critical information that hasn't been fleshed out in the spec/explainer etc that would likely be important to that process. The biggest being that I feel like we still don't have a clear scope here of how much the element is intending to do, which was one of the first concerns that we raised about the element in #55. The question of uses cases and alternatives raised by @DR-xR in #70 haven't really been fully addressed either; namely, we don't have clear differentiation between the element and a WebXR session and why a developer might prefer the element, and some of the reasoning that we do have for why to avoid AR is flawed (e.g. sites don't need the camera stream). One of the biggest things that also obviously still holds us back is that I don't believe I've seen any update to the spec with respect to declaring GLB as the minimum-required supported format, as discussed at TPAC.

As an additional point of order, I think there are two inaccuracies in the WHATWG issue:
1)

For privacy and security reasons, it's also not possible to integrate WebXR content with traditional DOM content, meaning that WebXR-based spatial experiences must provide all other interaction and GUI elements themselves.

Is incorrect, as the DOM Overlay spec does allow for that.

but provides significant benefits to web platform consistency on non-spatial devices as well.

Nothing that does on non-spatial devices today cannot be easily replicated in a privacy preserving way by existing libraries.

[zachernuk]

Good points Alex - on 1: while it is possible to merge DOM with webXR AR on phone, there's an implicit display surface there that we can't use in an HMD context to mix those responsibilities, where aims to pull the spatial content back into a majority DOM-governed presentation.

I agree on 2. that existing libraries can meet this need. My point is that using a single strategy for both spatial platforms and non-spatial ones can reduce author effort and increase consistency.

/agenda if nothing else, I'd love to discuss how much (or more likely, how little) we'll go into the details of <model> for WHATNOT on Thursday!

[alcooper91]

Good points Alex - on 1: while it is possible to merge DOM with webXR AR on phone, there's an implicit display surface there that we can't use in an HMD context to mix those responsibilities, where aims to pull the spatial content back into a majority DOM-governed presentation.

IIRC, @cabanier had discussed a few other options for putting DOM within the session (including e.g. summoning a full browser window). I know we had some privacy/security concerns, there but again it's not something that I think is impossible to do.

[zachernuk]

Leveraging some intentional consequence of visible-blurred

That's definitely a useful way to construct a spatial view of content while being able to present 2D material as well, though it still requires vending the 6DoF head-pose (and IPD) at framerate to the context. By placing the model content in the page, we begin the process of designing pages with spatial content, which I'm optimistic will let in more authors and allow for a greater breadth of immersive experiences.

[zachernuk]

@Yonet I would love to talk about this today, if only to raise that it's in the WHATNOT agenda for Thurs and if/how other folks want to speak to it at the time

[alcooper91]

+1 to chatting about it today.

Out of curiosity is the goal of the WHATWG issue to get advanced to a particular stage? https://whatwg.org/stages

[alcooper91]

Leveraging some intentional consequence of visible-blurred

Not sure where this quote came from?

By placing the model content in the page, we begin the process of designing pages with spatial content, which I'm optimistic will let in more authors and allow for a greater breadth of immersive experiences.

This remains the exact thing where I feel a tension between "Doing enough to actually be interesting/worth implementing" and "Shipping a game engine in the browser". It's a fine line we have to walk, and I don't think we've done a good job of drawing the (current) boundaries yet.

[zachernuk]

Not sure where this quote came from?

Sorry, I was interpreting your description of the suggestion @cabanier made here:

IIRC, @cabanier had discussed a few other options for putting DOM within the session (including e.g. summoning a full browser window).

(which is currently possible and a lot of fun to use, but wasn't an intentional way of leveraging the API as far as I knew)