miniSOG, developed as the first fully genetically encoded singlet oxygen photosensitiser, has found various applications in cell imaging and functional studies. Yet, miniSOG has suboptimal properties, including a low yield of singlet oxygen generation, which can nevertheless be improved tenfold upon blue light irradiation. In a previous study, we showed that this improvement was due to the photolysis of the miniSOG chromophore, flavin mononucleotide (FMN), into lumichrome, with concomitant removal of the phosphoribityl tail, thereby improving oxygen access to the alloxazine ring. We thus reasoned that a chromophore with a shorter tail would readily improve the photosensitizing properties of miniSOG. In this work, we show that the replacement of FMN by riboflavin (RF), which lacks the bulky phosphate group, significantly improves the singlet oxygen quantum yield ( Φ$_Δ$ ). We then proceeded to mutagenize the residues stabilizing the phosphate group of FMN to alter the chromophore specificity. We identified miniSOG-R57Q as a flavoprotein that selectively binds RF $in\ cellulo$ , with a modestly improved Φ$_Δ$ . Our results show that it is possible to modify the flavin specificity of a given flavoprotein, thus providing a new option to tune its photophysical properties, including those leading to photosensitization. We also determined the structure of miniSOG-Q103L, a mutant with a much increased Φ$_Δ$, which allowed us to postulate the existence of another access channel to FMN for molecular oxygen.