This paper tries to investigate that whether PHY/Application cross-layer optimization is really required for video transmission over next generation wireless networks? Or would a sequential allocation where optimization is independently performed at the PHY and Application layers work similarly? How does the cross-layer and non-cross-layer optimization perform also compared to the theoretical best allocation that one could apply, if the channel states and the user quality requirements were all known in advance? Is there a way to adapt to the channel variability? How do the unicast scenario extend to the multicast case? Given that a compromise in allocation must be found between the needs of all the users in the multicast group, it may be that cross-layer optimization is insufficient. Our numerical results show that cross-layer-optimization (XLO) significantly outperforms non-cross-layer-optimization (NXLO) for both unicast and multicast video transmissions, thereby pointing out the strong need for cross-layer solutions in video transmission.