We present a case study from the Kwanza basin offshore Angola where we built a detailed tilted transversely isotropic (TTI) model over an area of more than 8000 km2 with very limited well control using different types of geological constraints. We incorporate general knowledge of the area and analysis of data anellepticity in the process, derive Thomsen’s δ in wells from a neighboring area, and use spatially variable ε and δ fields, honoring the variation of the anellipticity and the geometry of the provided horizon interpretation. During velocity update iterations, we use tomography with implicit and explicit geological constraints to speed-up the convergence and optimize the image. The results are compared against images produced with both a much simpler regional TTI model and a legacy isotropic model. We demonstrate that accounting for TTI in complex media is a prerequisite for producing geologically plausible and interpretable images, and that adding dataand interpretation-driven complexity in the TTI models improves further the imaging of pre-salt targets.