Starshades are large, precisely shaped occulting masks flown in space and positioned between a telescope and its target star. The function of the starshade is to block direct starlight so that any exoplanets orbiting the target star may be observed. Formation flying with meter-level accuracy—needed to keep the starshade aligned between the telescope and the star—has been demonstrated at Technology Readiness Level 5. With the starlight largely blocked by the starshade, a second problem is to keep the telescope accurately pointed at the star, in the face of disturbances. The feasibility of measuring both telescope pointing and starshade position simultaneously from the residual starlight reaching a pupil-plane camera within the telescope beam train is investigated. Three sensing approaches, combining two types of sensors and two types of image-processing algorithms, are compared. It is shown that the required measurement accuracy can be achieved both for starshade position and telescope pointing, with little impact on scientific observations and with no additional hardware. Overall, differences in accuracy, speed, and memory requirements lead to a preferred architecture.