It is more reliable to work with AFNI datasets. While
almost all programs will read and write NIfTI without
any issues, scripts can have problems. Also as Daniel
and I have been pondering, it may be more problematic
(at the moment) to deal with oblique data in NIfTI.

I suggest that you stick with using NIfTI as an exchange
format, when going between AFNI, SPM and FSL.

The basics of datasets are covered in some of the early
slides in the "Introduction and Overview of AFNI" class
handout at:

[afni.nimh.nih.gov]

The HTML or PDF version is probably worth looking at.
Some slides discussing AFNI datasets start on page 26.

But there are 2 basic points about using AFNI datasets.

1. When specifying input, you specify everything up to
the view (e.g. my_dataset+orig). The .HEAD or .BRIK
is unneeded, but can be used.

2. When specifying a prefix for output, the view is not
used (e.g. -prefix my_new_dataset). The program decides
on the new view and applies it (usually the same as that
of the input).

- rick