Re: <model> explainer update

Hello,

This part of the explanation confused me. Is this saying new Fetch and CORS
rules would be made so model source data becomes completely inaccessible?
I'm confused why Fetch / CORS are involved as a rendering consideration.

Is there any precedence for this  in the browser today?

<Point of confusion>

Privacy considerations:
Rendered <model> data is not exposed to / extractable by the page in this
proposal, so no tainting is required. We do expect this would require
extensions to Fetch (a new destination type), Content Security Policy (a
new policy directive), and likely a few others.

On Wed, Sep 18, 2024, 6:43 PM Brandel Zachernuk <brandel@apple.com> wrote:

> Hi everyone,
>
> I have an update to the model explainer as initially posted by us back in
> 2022! I would love for folks to read it ahead of TPAC so we have a chance
> to reacquaint everyone with what <model> is, what kind of challenges it
> faces, and some potential ways of addressing those problems.
>
> It’s *not* my intention to set anything in the explainer in stone, but to
> get enough detail in one place for people to agree / disagree with. If you
> get the chance between now and TPAC, please take a look through the
> document (and the demo!) to see the refinements to the proposal. There’s a
> PR out for both to land in main, but you can read from the branch here:
>
> https://github.com/immersive-web/model-element/blob/explainer_demo/explainer.md
> Additionally, there is an interactive explainer demo that covers the broad
> context that <model> is attempting to cover- it’s also in the PR to get
> “officially” hosted, but I have a copy of it up here:
> https://zachernuk.neocities.org/2024/model_explainer/
> It’s intended to be viewed by scrolling on a computer, but should work on
> a phone as well.
>
> I look forward to further discussion on all of this next week!
> Many thanks,
> Brandel
>
> *What’s new:**Tighter limitations on v1 functionality*As an initial
> candidate, I feel like <model> shouldn’t encode interactive state beyond
> that presented via an animation timeline or camera controls.
> *Update to camera / view controls*The first proposal’s use of “camera”
> feels like it was too limited. entityTransform supports a DOM-native API
> for constructing a more sophisticated view without adding too much extra
> complexity. The API is also synchronous (even if the view updates are
> managed out-of-process) - more about that below.
> *Bounding box information*Used in service of understanding how to set up
> a view of a model, in the event that the contents aren’t known by the
> author already.*environmentmap attribute and events*Important for an
> adequate level of visualization control, I feel like applying one IBL is a
> “good-enough” start to allow authors to present model media in different
> contexts - and means model *files* are not impairing their ability to
> display correctly in a mixed-reality view.
> *Reducing the use of Promises*While many APIs like HTMLMediaElement and
> webGL/webGPU offload activity to other processes, that’s not always a good
> reason to use a Promise-based API. Particularly with attributes like view
> pose and animation time, a reasonable answer *now* is likely better than
> a perfect answer at some unknown time in the future.
> *Manipulation terminology change*The discussed automatic controls were
> previously known as “interactive’ mode, which was noted to be an ambiguous
> term. Clarifying that it’s related less to a stateful interactivity and
> more to a mode of interaction on a “stage” in an “orbit” mod is intended to
> differentiate this.
> *Removal of audio*While audio is part of some 3D model asset types, it’s
> a capability that is already available through existing APIs, so it’s not a
> critical part of an initial implementation to provide that functionality.
>

Received on Thursday, 19 September 2024 02:38:01 UTC