Thanks ziad/Daniel. About the blurring (to ziad), when I look at a raw and a deobliqued dataset, there are significant differences in the time courses (and hence, spatial maps too). If deoblique-ing does a nearest neighbor interpolation, then technically it may not be blurring but instead making it blocky, but that I feel is worse. In either case, since I need to only align datasets within a run to each other, 3dvolreg will work fine.