The process outlined here is very similar to how Simfox's renderer works to my knowledge. I can only see one difference, that this method still uses standard Z5 resolution textures (128x128), just maps them over a smaller area of the model whereas AFAIK Simfox's creates the Z5 textures at double the resolution and maps them onto a standard LOD.
You're not rendering a larger texture at all, you're creating more of them and reducing the size of the area on the object each one is mapped to.
This particular method will also have the downside of producing models with a higher polygon count than using Simfox's renderer, but still using the same (Or more) space for Texture storage.
Upside is this is also backwards compatible with the original BAT.