I have oblique EPIs and non-oblique anatomicals.
I followed the procedure recommended in this thread:

1. 3dWarp run1 with newgrid
3dWarp -deoblique -verb -newgrid 3 -prefix run1_dob run1+orig.

2. 3dWarp run2 with gridset
3dWarp -deoblique -verb -gridset run1_dob+orig. -prefix run2_dob run2+orig.

to look at how different these datasets were, i ran a 3doutcount it, and it's Really different -- the number of outliers at a given TR differ by an order of magnitude.

Question1: is that ok? what does that mean?
Question2: if my steps look correct + if it's ok for the warping to alter the data that much, is it correct to assume that the 3dWarp should be the absolute first thing done to the data, before ANY other preprocessing steps?

Thanks for advice,
Dharol Tankersley