Hi Torsten,

The obliqueness does not matter too much here, but the
data grid does. We suggest going to standard space during
the volreg block of single subject processing, and also to use
a grid that stays very close the original voxel sizes.

Consider what afni_proc.py does here. Note that after the
regression, the FWHM is calculated from the errts data.

s12.proc.FT.align

If your group data is on a higher resolution grid than the
single subject data, then that is where 3dFWHMx and 3dClustsim
should be run. Though again, it is good to stay close to your
original voxel resolution.

- rick