You are correct in that my aim was to target multiple different platforms from desktops to tablets to mobiles, from the benchmarks I have run on some devices I will need to do some further optimizations to make it work and scale across mobiles.
WebGL is a technology I am playing with and it will be the next feature I develop after the audio stuff I'm working on, the thing that bugs me about WebGL is that, unlike the 2D canvas context, WebGL is not part of the HTML5 spec and is unlikely to become widely adopted for a while if ever (
looking at you Microsoft). Heck some browsers that do support WebGL have it disabled by default... what's up with that (
looking at you Opera and Safari)?
three.js is a very impressive solution and I have much respect for the team behind it. I will be looking at using it for the WebGL stuff, however my motivation is to create a simpler and easier to use engine and associated HTML5 based tool-set, geared specifically towards the creation of casual games and game prototypes quickly and easily.
However a game engine is much more than just rendering... we are talking audio engine, physics engine, input, networking, AI, editors, scripting, resource management, packaging, deployment and so on.
The last but probably most important reason is... for fun