From what I've seen, init() is more common for functions that don't allocate, i.e. return a plain struct, and create() for ones that do allocate and return a pointer.
#Naming conventions - When to use init() and when create()
raven matrix
bright lichen
There's no real hard rule about it, name your methods whatever makes the most sense. Another common convention is for a type not to have an init function, but instead to exist alongside a function of the same name, which can be nice for allowing more type inference, e.g., take std.io.fixedBufferStream, which returns a std.io.FixedBufferStream(T). Or sometimes types just don't have initialisation functions, because the process to initialise them is trivial, e.g. like std.Thread.Mutex, or std.ArrayListUnmanaged(T).