Until now, I have not tried to implement any of the methods proposed in the papers mentioned above. Most of them call for finite-difference algorithms, and it appears to me necessary to find a way to produce 3-D grids easily and rapidly from a 3-D model.
The Eikonal equation for time migration introduced by V.J. Khare and the theory that comes with it seem to represent an interesting way to obtain a de-migration process accurate enough to handle most of the complicated models.