I don't know much about Superpowers internals, but I don't understand why Superpowers must manually parse the models when Three.js already has a bunch of importers ? Is it because you need some data about the model that Three.js doesn't provide ? Also, another advantage of using the Three.js loaders is that you would have out of the box support for all the other formats that Three.js supports (Collada for example).

Thanks :)