What is the orientation of files in the anat output directory?

[civier]

I assume it is the dwi data orientation. Could you confirm?

[Lestropie]

T1w and derived images retain the voxel grid of the original T1w image. As of 0.5.0 it's actually the DWI data that are transformed to overlap the T1w information, not the other way around as it was previously (since the T1w is more likely to be used as the central reference in multi-tool / multi-modality workflows). But that's a purely rigid-body translation that is applied to the image header; no re-gridding takes place.

So there's no guarantee of common orientation between the T1w and DWI image axes. The only guarantee is that upon selecting any arbitrary location in scanner space, and independently transforming that location into voxel coordinates in the T1w and DWI data, you should be sampling image data from the same biological location. This is fine for tractography since sampling data from the FOD image and the 5TT image are entirely independent, so there's no issue with the scanner2voxel transformations being different.

[civier]

Thanks for the explanation. Very useful.
But ... I'm actually still using 0.4.2 :-) How do things work there?

[Lestropie]

T1w was transformed to align with DWI rather than the other way around; but still only a rigid-body transformation altering the image header.