Layout migration has re-emerged as an important task due to the increasing use of library hard intellectual properties. While recent advances of migration tools have accommodated new metrics, the underlying engine is based on the one-dimensional (1-D) layout compaction algorithm, largely due to its efficiency compared to its two-dimensional (2-D) counterpart. In this paper, we propose a new method that can overcome the artificial constraints introduced by the 1-D compaction algorithm, thereby effectively achieving the quality of 2-D compaction, yet keeping the computational cost almost as low as 1-D compaction. Our method is based on the application of soft constraints, or artificial constraints that are initially relaxed, and gradually tightened to be satisfied. We demonstrate the effectiveness of our approach by successfully solving the difficult 1-D compaction instances we found in the migration of Berkeley low power library, originally developed for 1.2um MOSIS process, into TSMC 0.25um and 0.18um technology.