Setting the viewpoint

The viewpoint is set using the estimated orientation $\hat{q}[k]$, although it might need to be adjusted to account for alternative timings, for the purpose of prediction or image warping, as discussed in Section 7.4. Let $\hat{q}(t)$ denote the estimated orientation for time $t$. In terms of the transformations from Section 3.4, we have just estimated $R_{eye}$. To calculate the correct viewpoint, the inverse is needed. Thus, $\hat{q}^{-1}(t)$ would correctly transform models to take the estimated viewpoint into account.