Both. Some of the stuff we needed was thankfully added in Visual Studio 2013. Stuff like constexpr and noexcept has special compiler switches, some stuff is just flat out our own due to missing support (fixed width types).
The biggest problem honestly was that some things were supposed to be implemented, but then we ended up running into some strange bug... Especially when it came to rvalues, move semantics and (perfect-)forwarding. Oh dear, that was fun.

There's no way to get it running on VS2012 due to missing core language support, but for VS2013 we could work around every obstacle that was left.
That means that, if you want to use Rayne, you'll have to switch to VS2013 as well because of ABI constraints and header that require it, but that shouldn't be that much of a problem I guess.

Edit3: I would love to actually go ahead and switch to C++14/C++1y. It seriously does allow for some fucking neato stuff, and the support would be there in Clang. Function lifting thanks to function return type deduction... Man, that would be sweet as fuck (come to think of it, maybe you can actually do something with decltype() and function templates). But yeah, the future for C++ looks amazing, and C++11 definitely feels like a whole new language. I wish adoption and support for new standards would be much faster, though adoption is inherently tied to support.