Direct Energy Deposition (DED) Additive Manufacturing is Well Suited to Fabricating Large Thin-Walled Metal Structures Such as Rocket Nozzles but Suffers from Layer-To-Layer Defect Propagation. Propagating Defects May Exhibit as Slumping or a Ripple in Bead Geometry. Recent Works Have Used Repetitive Process Control (RPC) Methods for Additive Manufacturing to Stabilize the Layer-Wise Defect Propagation, But These Methods Require Repetition of the Same Path. However, Typical Thin-Wall DED Applications, Sometimes Referred to as Vase Structures, Have Changing Paths with Each Layer Such as Expanding or Contracting Diameters and Changing Profiles. This Paper Presents an Extension to Optimal RPC that Uses a Geometric Mapping Method in the Learning Algorithm to Project Previous Layer Defects onto the Current Layer, Even When Paths Are of Differing Profile and Length. the Novel Method is Implemented on a DED System and Sample Parts with Layer-Changing Geometry Are Printed. the Experimental Results Demonstrate that the Method is Capable of Stabilizing the Layer-To-Layer Ripple Instability and Producing Parts of Good Quality.