[Insight-users] Basic question on interpolation

[motes motes]

Have a look at the below image:

http://img6.imageshack.us/img6/9570/interpoll.jpg


I am going to move the white pixel based on the deformation vector in
the black dot. Now my question is;

1) Should I do it for each dimension separately => new location (6,11) or

2) Should I multiply the offset in each dimesion and then use the
result as a single weight for the deformation vector => new location
(18, 11)

Are there anywhere its possible to read why 1) or 2) above is correct?