Hi Peter,
I have 2 EPI scans (which may or may not have slightly differnet oblique angles from each other) and 1 T1 scan which has a different angle. I am using freesurfer for segmentaiton, skull stripping and normalization (not surface mapping at this time). I want to only have to process the freesurfer files once and then align the brain.mgz T1 and corresponding masks to the EPI scan as opposed to aligning the EPI scans to the anatomical, as I want to minimize interpolation of the EPI data. I used bbregister in the past as part of my pipeline for scans taken with no oblique angles and it worked perfectly when viewing in AFNI in the past. Now that the T1 and EPI are taken obliquily and 2 different angles I am running problems with it being overlayed in AFNI, which I would use for the EPI processing.

I decided to then try afni_proc.py/3dAllineate for registration step and so I was getting the error messages which is why I replied to this old post, to see what the issues are and how to resolve them. Daniel suggested running 3dWarp -oblique2card on the original data when he thought it was only 1 EPI and 1 ana tomical scan, however with 2 EPI scans I would have to run 3dWarp twice and process 2 different freesurfer threads per subject (since EPI are at different angles potentially), which is why I wanted to figure out how bbregister and 3dAllineate stores transforms to NIFTI so that I can fix the issue after freesurfer processing. Ideally I would like to use bbregister (since it uses a boundary based cost function for alignment) and figure out how to write the transform to the NIFTI header while properly handling the obliquity issues so AFNI can read it properly. However, if this was not possible, than I can look into the scripts you mentioned above to see if it allows for alignment of anatomical to EPI (last time point), as I do not want to align the EPI to anatomicals. Hope this clarifies what I am trying to accomplish/questions I was asking.

Thanks,
Ajay