Tckgen: What’s the difference between -seed_gmwmi and -seed_dynamic? Which one should be used for HCP data?

Hi Community,
In the command tckgen, what’s the difference between -seed_gmwmi and -seed_dynamic? Now I have a HCP data, which one should I use? I have run them and calculated fiber lengths, as follows

tckgen WM_FODs.mif 1M.tck -act 5TT.mif -backtrack -crop_at_gmwmi -seed_dynamic WM_FODs.mif -maxlength 125 -select 1M -cutoff 0.06

tckgen WM_FODs.mif 1M_gmwmi.tck -act 5TT.mif -backtrack -crop_at_gmwmi -seed_gmwmi gmwmi.mif -maxlength 125 -select 1M -cutoff 0.06

Any suggestions there?
Cheers,
Jian

The GM/WM interface seeding was described in the original SIFT paper, and shown to provide streamline densities with better representation from shorter, more superficial pathways. It was subsequently assessed more thoroughly in this paper. Dynamic seeding was described in the subsequent SIFT2 paper, and shown to do better again on that front. It’s inspired by the ideas behind SIFT, and aims to compensate for seeding-related biases in tractography by preferentially seeding from low streamline density areas. I recommend you look through the papers for details, and if you still have specific questions after that, I’m sure @rsmith will be able to give better answers than me…