Any seismic inversion that delivers a solution containing frequencies outside of the seismic bandwidth does so by using additional information. This information is not contained in the seismic data. Examples of such information include low frequency models, assumptions in algorithms (sparse spike for example), assumptions about spatial variability (variograms), and existing well control. When we apply these constraints, we are really implying that we know something about the geology that will give a more useful answer if it is incorporated. With so many advanced techniques for inversion, it is easy to lose sight of the geological meaning of the mathematical constraints. Ideally, a constraint should be geologically accurate and bias the answer minimally. In this presentation, we will show how we can take advantage of large separation of the elastic properties of lithofacies, to achieve remarkable accuracy of inversion results from a VSP corridor stack used to look ahead of the bit.