We introduce a Progressive Positional Encoding (PPE) layer, which gradually exposes signals with increasing frequencies throughout the neural optimization. In this paper, we show the competence of the PPE layer for mesh transfer and its advantages compared to contemporary surface mapping techniques. Our approach is simple and requires little user guidance. Most importantly, our technique is a parameterization-free method, and thus applicable to a variety of target shape representations, including point clouds, polygon soups, and non-manifold meshes. We demonstrate that the transferred meshing remains faithful to the source mesh design characteristics, and at the same time fits the target geometry well.