Hello Hanchuan,

you're right, but unless I missed something, the landmark outputs are only created for warping if landmarks are given (i.e., when using -w 10). There is no output for the affine global transformation (-w 0). Normally, I do my registrations in two steps, first computing the global transformation and producing the image result, then computing the warped image. Like this : Or maybe it is possible to do both transformations in one round with another parameter for -w ? In this case, it saves both transformations in a file ?

- Another reason why I would like to use a displacement field, is that in one channel I want to detect nuclei (anti beta galactosidase staining). The solution I found is to detect nuclei in the original channels, then compute the transformation with the reference channel, then compute the new centers of mass of the nuclei with the displacement field (I only translate the centers of mass, keeping the same diameters). I would like to avoid counting nuclei after transformation, because my images look already a bit blurry in dense nuclei regions.

Thanks for your help
Best Regards

Thomas