Just to avoid confusion, perhaps it should be noted somewhere (here? :))
that camera location must be the same for all images. Otherwise the detail
views could show something that wouldn't be visible in the main image when
rendered at higher resolution; and if that would be OK, then why not render
three completely different views of a complex model, where none of the
images is really a closeup of another.
I'm not sure if this isn't just nit-picking, but I had to think about it
when I was trying my detail views :)