Something is still missing here. I have another dataset, test2+tlrc, which is already Talairach'd and exists within the same volume of space as norm_warped_axial+tlrc and warped_axial+tlrc (according to 3dinfo). I can supply it if you want- it's the output of exactly the process you suggest, if I'm understanding you correctly. It, too, is visible when the anatomy is warped_axial but not when it is norm_warped_axial.
This is annoying because my goal is to produce Taliarach'd functional datasets from +orig functional datasets using AFNI's "adwarp", with a Taliarach'd anatomical dataset compatible with our software as the geometry prototype. What appears to be happening when I try to do this is that I get a lovely dataset of all zero's when I use a dataset in the format of norm_warped_axial+tlrc as the prototype, but a perfectly fine one when I use warped_axial+tlrc as the prototype. warped_axial and norm_warped_axial are structurally identical, but one uses the RAI orientation and the other uses RAS (which is compatible with our coordinate system).
-Joel