The paper introduces a novel direct sparse visual odometry (DSO) formulation that combines a fully direct probabilistic model with consistent, joint optimization of all model parameters, including geometry and camera motion. This approach is achieved in real-time by omitting the smoothness prior used in other direct methods and instead sampling pixels evenly throughout the images. The method does not rely on keypoint detectors or descriptors, allowing it to sample pixels from all image regions with intensity gradients, including edges and weak intensity variations on white walls. The model integrates a full photometric calibration, accounting for exposure time, lens vignetting, and non-linear response functions. Extensive evaluations on three datasets show that the proposed method significantly outperforms state-of-the-art direct and indirect methods in terms of tracking accuracy and robustness. The paper also discusses the motivation behind the direct sparse formulation, the contributions, and the detailed implementation, including the front-end for data selection and initialization.The paper introduces a novel direct sparse visual odometry (DSO) formulation that combines a fully direct probabilistic model with consistent, joint optimization of all model parameters, including geometry and camera motion. This approach is achieved in real-time by omitting the smoothness prior used in other direct methods and instead sampling pixels evenly throughout the images. The method does not rely on keypoint detectors or descriptors, allowing it to sample pixels from all image regions with intensity gradients, including edges and weak intensity variations on white walls. The model integrates a full photometric calibration, accounting for exposure time, lens vignetting, and non-linear response functions. Extensive evaluations on three datasets show that the proposed method significantly outperforms state-of-the-art direct and indirect methods in terms of tracking accuracy and robustness. The paper also discusses the motivation behind the direct sparse formulation, the contributions, and the detailed implementation, including the front-end for data selection and initialization.