Natural photoactive proteins serve both as inspiration and as templates for efforts to design new proteins with photofunctions beyond the biological context of the template. Inevitably, changing the amino acid composition will modulate properties, but there is no systematic understanding of how the desired excited-state control can be achieved nor how modifiable a given template can be. One example is reversibly photoswitchable fluorescent proteins that are key actors in enabling bio imaging beyond the optical diffraction limit. Yet, their photoisomerization quantum yields are consistently low, hampering their use in active control applications. Understanding the origin of these limitations could facilitate breaking free of the photo functional optimum of the template. In this talk, I will discuss our recent efforts to unravel photoisomerization bottlenecks in negative reversibly photo switchable fluorescent proteins using an arsenal of computational tools.