We investigate the potential energy surface and dynamics of atoms impinging upon pyramidal defects on fcc (100) and (111) surfaces. Illustrative molecular dynamics trajectory results are used to highlight the subtleties of the growth processes. Homoatomic systems of both Cu and Pd are considered in order to determine the influence of extreme differences in bonding coordination variation, i.e. the binding energy per atom decreases by a factor of 3.3 from Cu(bulk) to Cu 2 but more dramatically by a factor of 7.5 for Pd(bulk) to Pd 2. We find that stable adsorption sites on pyramidal facets only exist for three layer high (and higher) facets, but even these sites are not as energetically favorable as those at the pyramid's base. This single energetic effect can promote a Stranski-Krastanov type growth pattern, with a critical roughness of no less than three layers being required before a 3D growth mechanism can propagate in a surface temperature regime where post impact diffusion is small compared to the deposition rate. The transition to 3D growth at the minimum possible roughness would be found for Pd on the (100) face where sticking to the sides of even the three layer pyramid is a high probability event, at a low enough temperature to make post-impact diffusion negligible. However, for the Cu system, we find that the “downward funneling” mechanism smoothed the growth for the three layer pyramid and is still important for the five layer pyramid, indicating that Cu will exhibit layer-by-layer like growth even past three layers. A comparison of scattering simulations from defects show that the growth mode is diffusion limited on the fcc (100) crystal face, but becomes dominated by impact induced disruption of the original structure for a (111) surface. This implies that the (111) surface will grow more smoothly as a new atom displaces pre-adsorbed atoms more easily, an effect that is not included in the “downward funneling” model. The ratio of the diatomic molecule's binding energy to the bulk cohesive energy is shown to provide good insight into the preferred growth mechanism. As the ratio increases, at least in the range spanned by typical metals, the film grows more layer-by-layer like on the (100) surface and less layer-by-layer like on the (111) surface. On the (100) surface, stronger low coordination bonding makes the defect structures less able to adsorb the impacting atom on the sloped faces, increasing the effectiveness of “downward funneling.” On the (111) surface, stronger low coordination bonding inhibits impact disruption of the defects, decreasing this important smoothing mechanism.