@Joey10582@Scobleizer@cly3d@AdaRoseCannon There are a lot of features in USD that are not in glTF (for good reason). It's not a competition where only one format will win.
Also, Apple is implementing WebXR, in Open Source.
@kmcurry@mrdoob @TanayDimri @SebAaltonen Honestly, I'm not sure! There are lots of great features in USD that would be useful on the Web. But, at the same time, it's heavy (as you mentioned) and infinitely extensible (which is sometimes good, but makes interoperability hard).
@mrdoob@smfr@alexanderdanilo There has been a lot of investigation/documentation into how to get good performance out of 2D transforms. 3D (as @smfr said, really 2.5D) should be the same.
I admit that you see a lot of bad designs that use 3D transforms. There are also some complications with CSS layout.
I can't find the tweet now, but someone suggested <model> was a way to own the Metaverse.
1. If you wanted to own something why would you try to develop it as an open standard that works with other technologies?
2. I don't know what the Metaverse is.
@mrdoob@alexanderdanilo I can answer this one!! CSS Transforms was a way to give a very simple CSS-based way to do compositing in Web layouts (i.e. accelerated rendering). Once you do that, the 3D implementation comes basically free, and allows for some nice designs (with care).
@arturitu@mrdoob Are you saying we'd want to turn off canvas? No way. <model> isn't trying to compete with canvas at all. Canvas will always be more powerful - a black box that allows you to do anything!
@mrdoob @TanayDimri @SebAaltonen Yep. USD needs a lot of improvements to be well-suited for the Web. It's very popular as a powerful interchange format/API between DCCs, but not as good for final mile delivery to a browser.
@mrmaxm@mrdoob It's definitely the hardest part - how to expose an API that isn't overwhelming (both to developers and the specification), balanced with giving something useful.
(The other hardest part is compatible rendering)
@mrdoob@TheJare That's true in the initial proposal, but it's about taking small incremental steps.
Step 1: Plonk a 3d object in a page
Step 2: Expose some API to control cameras, animations, and maybe register event handlers
Step 3: Expose the scene graph
@warrenm@mrdoob Hi Warren! I'd like to hear more about what level of abstraction you think would be right. We've tried to explain why we aim at this level, but probably haven't done a good job.