@CreamFilling512: It was tried with VRML already. I personally think the WebGL approach is better, because it represents how people ACTUALLY write 3D apps and doesn't add multiple layers of abstraction in the picture.