Visual odometry is a key component of autonomous vehicle navigation due to its cost-effectiveness and efficiency. However, it faces challenges in low-light conditions because it relies solely on visual features. Tomitigate this issue, various methods have been proposed, including sensor fusion with LiDAR, multi-camerasystems, and deep learning models based on optical flow and geometric bundle adjustment. While theseapproaches show potential, they are often computationally intensive, perform inconsistently under differentlighting conditions, and require extensive parameter tuning. This paper evaluates the impact of image enhancement models on visual odometry estimation in low-light scenarios. We assess odometry performance onimages processed with gamma transformation and four deep learning models: RetinexFormer, MAXIM, MIRNet, and KinD++. These enhanced images were tested using two odometry estimation techniques: TartanVOand Selective VIO. Our findings highlight the importance of models that enhance odometry-specific featuresrather than merely increasing image brightness. Additionally, the results suggest that improving odometryaccuracy requires image-processing models tailored to the specific needs of odometry estimation. Furthermore, since different odometry models operate on distinct principles, the same image-processing techniquemay yield varying results across different models.