The case where applying -weight might be
useful is if there is a high intensity difference
between the base and input volumes. For
example, when running 3dvolreg on Monkey
data, the high intensity muscles around the
jaw have a big effect on registration. Using
-weight with a brain masked volume (not
binary necessarily) can restrict the result
to the actual brain region, giving much better
results (based on the brain, rather than the
jaw muscles).
- rick