To deeply reveal the deformation evolution mechanism of rocks under uniaxial compression conditions, and to study the evolution characteristics and patterns of their deformation paths before failure, red sandstone was taken as the research object. The digital speckle correlation method was employed for full-field deformation monitoring, and uniaxial compression experiments were conducted to obtain the evolution characteristics and patterns of the full-process deformation paths of red sandstone under uniaxial compression. The results show that initial loading stage: the horizontal deformation increment paths of rock sampling points remain essentially consistent, while the horizontal deformation direction undergoes systematic alteration during loading. Advanced loading stage: significant differentiation emerges in both vertical and horizontal deformations, with deformation paths exhibiting an umbrella-shaped divergence distribution pattern as loading progressed. Horizontal deformation increment values demonstrate stage-specific variations across different loading periods. The differentiation of sampling point deformation paths shows direct correlation with their spatial positions and crack formation locations. Throughout the complete loading failure process, deformation path angles exhibit distinct oscillatory characteristics: initial stage fluctuations displayed large amplitudes with long cycles, intermediate stage fluctuations show small amplitudes with short cycles, and final stage fluctuations manifest large amplitudes with short cycles. Particularly around peak load conditions, the angular deformation path curve demonstrates abrupt increments and high-frequency oscillations with fluctuation amplitudes reaching ±180°. The evolutionary characteristics of deformation paths in red sandstone under uniaxial compression effectively reflect the stress state and deformation failure trends throughout the loading process, providing valuable references for understanding rock failure mechanisms and conducting short-term rock failure prediction research.