A pixel-based inversion approach is introduced for interpretation of deep directional resistivity logging-while-drilling electromagnetic measurements. It can be used for real-time reservoir mapping while drilling to accurately place the well with respect to reservoir boundaries and fluid contacts as well as for reservoir characterization in high angle and horizontal wells. The methodology is based on the Gauss-Newton approach, using an adaptive l1-norm regularization and a robust error term to determine the 1-D formation anisotropic resistivity profile and local formation dip. The new inversion allows high-resolution imaging of the subsurface resistivity distributions and consistent integration of measurements of different depth of investigation. To further enhance the robustness and accuracy of the subsurface images, a fast approximate 2-D inversion is developed, which uses approximate modeling based on locally 1-D models, resulting in reservoir images of improved accuracy and consistency in mildly 2-D and 3-D scenarios. The inversion's ability to generate a stable robust but detailed resistivity image of the formation near the wellbore is proven on synthetic and field data.