We employed a novel combination of digital image correlation (DIC) and grain-based hybrid finite–discrete element method (GB-FDEM) to improve the comprehension of the relationships between microstructural features and the mechanical properties of granitic rocks. DIC and numerical results showed that macrocracks initiated and propagated along grain boundaries among different minerals driven by the high stiffness contrast between the compliant biotite and the stiffer feldspar/quartz grains. Surface deformation analyses revealed that tensile-dominated macrocracks open at monotonically increased rates before the crack damage threshold, and the opening accelerated afterwards with the increased shear component. The onset of the acceleration of the opening rate of macrocracks can be used to infer the crack damage threshold. Both strain and acoustic emission were used to infer damage stress thresholds in the synthetic numerical samples. Numerical results showed that the damage stress thresholds and uniaxial compressive strength decrease with increasing grain size following log-linear relations. Coarse-grained samples tend to fail by axial splitting, while fine-grained samples fail by shear zone formation. Biotite and quartz contents significantly affect mechanical properties, while quartz to feldspar ratio is positively related to the mechanical properties. Our study demonstrates the capacities of DIC and GB-FDEM in inferring damage conditions in granitic rocks and clarifies the microstructural control of the macroscopic mechanical behaviors. Our results also provide a comprehensive understanding of the systematics of strain localization, crack development, and acoustic emission during the rock progressive failure process.