The peak of SPMG1 is 0.175441 ~= 1/5.7. So theoretically, all of the betas with SPMG1 will be 5.7 times those you would get with SPMG1(0).
That scaling will not have any effect on the group statistics, aside from computational truncation differences. It should only scale the group betas by that same 5.7.
GAM will have a real difference, as that curve has a different shape.
- rick