Patellofemoral cartilage deformation and contact area changes in response to in situ loading were measured with high-resolution MRI. In situ loading was realized with a pneumatic loading device and motion artifacts were corrected with prospective motion correction based on optical tracking of the knee cap. Semi-automatic cartilage segmentation based on deep learning proved essential for robust quantification of the load-induced changes. Cartilage thickness and contact area showed significant and weight-dependent changes in response to loading. The patellofemoral deformation and contact mechanism under loading might be used for investigation of the knee biomechanics and as a biomarker of early-stage cartilage degeneration.