We present a new extension and application of prior work on LDDMM (Large Deformation Diffeomorphic Metric Mapping) flow matching to generative modeling for uncertainty quantification in the context of blood vessel geometries estimated from medical images. Conditional LDDMM flow matching is used to generate arrays of variants of existing aorta geometries; simulations of pulsatile flow on these are performed and the results compared in terms of medically relevant biomarkers.