The eye provides a window through which a complete vascular system of arteries, capillaries, and veins can be directly observed. An adaptive optics scanning laser ophthalmoscope (AOSLO), a custom-built optical microscope for the living human eye, can be used to directly acquire videos of blood flow through the smallest capillaries in the eye. However, in the absence of invasive contrast agents, the contrast of blood cells and capillaries is very low. Moreover, it is difficult to determine the locations of all capillaries, and therefore tracking and speed quantification of blood cells is hindered. In human eyes, contrast agents such as fluorescein are routinely used only in the later stages of certain diseases; as with any invasive procedure, there is also a risk for adverse side effects. Instead of injected contrast agents, we used intrinsic signals from moving blood cells to create contrast. By applying custom motion contrast enhancement methods to AOSLO videos, we were able to noninvasively visualize complete capillary networks as well as characterize hemodynamics through the capillaries in the eye. We investigate capillaries in healthy and diseased eyes, and show that the flow dynamics of leukocytes and plasma gaps are heterogeneously distributed, with certain capillaries accounting for a clear majority of leukocyte traffic, and other capillaries primarily featuring plasma gap flow. Such capillaries may serve specific functional roles, such as to prevent inactivated leukocytes from entering exchange capillaries, or to serve as relief valves to minimize flow disruption due to the presence of a leukocyte in a neighboring capillary. In diabetes, we found evidence of capillary remodeling even before conventional clinical methods detected any changes. We establish that motion signals can be used to generate intrinsic contrast for visualization and analysis of capillaries and blood cells. These methods are important for evaluating diseases that affect the microcirculation, such as diabetes.