Hi, Clément-

To be clear---I think "-volreg_tlrc_warp" takes no additional arguments, it is just a switch or flag.

When using @animal_warper or @SSwarper, you likely have the full warp in 2 pieces (an affine part in a *.aff12.1D file, and a nonlinear part, in a *_WARP* volumetric dset).

The afni_proc.py option "-tlrc_NL_warped_dsets" does require using 3 arguments after it: an anatomical datset, the affine *.aff12.1D file, and the warp *_WARP* dset. I don't see a way to *not* provide all three.
+ If there is an *.aff12.1D file, I think you would always want to include it to have the correct warp (it would need to be concatenated internally with the nonlinear warp).
+ If you calculated a warp differently, and you *don't* have a linear affine part to concatenate---that is, the WARP dataset is the *full* and *complete* warp, then perhaps you could fake an affine transformation that would be reasonable. Namely, make a file, say identity.aff12.1D, containing an "identity" affine transform:

1 0 0 0 0 1 0 0 0 0 1 0

... and provide that as an argument where the affine should go.

How does that work?

--pt