Hi, Sorry - I'm not quite sure what you are asking here. Are you saying that you have a method that works well for poor 3D images, but with good 3D images it fails? I'm also not sure what your script actually does. If you can let me know a bit more then hopefully I can help. All the best, Mark Joerg Magerkurth wrote: >Hi Mark, > >sorry that I answering so late, I was busy with other studies. Now I have >registrate the most of the subjects and the biggest part of it looks good. > >I have changed your routine, such a way, that the script do an iterative way >of four steps: every step take the matrix from the last for the init-option. >So the result becomes still a little bit better in each step. I have tried >this with good 3D datas, but there it goes wrong. I would be gratefull if >you can explain this. But it is not so important. Only if you have some time. > >Best, > > Joerg > >