The controller synthesis problem for CPS is often reduced to solving omega-regular games with an optional optimization criterion. The criteria commonly used in the literature on omega-regular games are however frequently unsuitable for obtaining high-quality CPS controllers as they are unable to capture many, if not most, real-world optimization objectives. We survey a few such cases and show that such problems can be overcome with more sophisticated optimization criteria. The synthesis problem for them gives rise to ugly games, i.e. games that have complicated definitions but relatively simple solutions.