Processing gesture input includes obtaining hand tracking data for a hand from one or more sensors, wherein the hand performs a pinch in the hand tracking data, determining a hand position and a hand orientation for the hand from the hand tracking data, applying a predefined offset from the hand position based on the hand orientation to determine a pinch centroid, and determining an input location based on the pinch centroid. The pinch centroid is used in an indirect motion mapping pipeline to determine characteristics of user input based on user motion.