Motion of one or more physical objects relative to a display surface of a display system is detected and an optical flow determined from the motion is used to manipulate a graphical object presented on the display surface. The one or more physical objects are detected in response to light reflected from the object(s) and received by a video camera. The optical flow is determined from the video camera image by identifying flow vectors for points in one or more patches included in the image that correspond to the physical objects. A proximity of a physical object to the display surface can be determined based on an intensity of light reflected from the physical object(s), or using a touch sensor such as a capacitance, pressure, or electromagnetic sensor or the like. Based on the optical flow, the graphical object can be translated, rotated, and/or scaled in size.