I was checking how the site for the Apple Vision Pro was made. Can confirm that this section was made with Three.js. They're using a pretty clever technique that breaks the Vision Pro model in segments using GLTFs and it's reassembled on the browser. Pretty darn cool.

@Apple @threejs That technique is common and is exactly what we did for

@Apple @threejs is this more performant than simply hiding/showing different meshes within the glTF jsx group? Assuming this is r3f... @0xca0a @Cody_J_Bennett thoughts?

@Apple @threejs @0xca0a @Cody_J_Bennett They don’t use React. Is a complex @threejs app, with a bunch of shaders to blend/fade-in different scenes. They made like a camera control system driven by JSON assets that contains all the choreography already pre-calculated.

@Apple @threejs Wait, what? They are using GLTF instead of their lovely USDZ? 😮

@Apple @threejs Perhaps the people from @apple who actually built this share more with us ❤️

@Apple @threejs I think it's just one gltf changing camera angles and at some point changing a stencilRef that allows the back band to fade when viewing from the front.

@Apple @threejs This is the usual flight of the camera and the deployment of details according to the transparency of the material. There is nothing complicated here, the model is well made and the lighting is selected using hdri

@Apple @threejs Mmm, but did you noticed that you’re doing arbitrary interpolations between different camera targets, right?

@Apple @threejs I can make a torus rotate a bit.

@Apple @threejs Nice find, any tutorials or articles that explain this technique more in depth?