The rapid adoption of distance learning means that significant lecture video content is available on-line. However, access from mobile devices is hampered by low bandwidth and small screen size. In this paper, we address these issues by manipulating two key elements of lecture video-displayed slides and laser pointer gestures. Displayed slides need to be very crisp compared to background content. Fortunately, the needed data is available from the presentation slides, and we describe a method for splicing them into the video on the client side, increasing fidelity, and reducing bandwidth needs. This operation removes laser pointer gestures, which are often lost due to compression, and are hard to see on the small screens of mobile regardless. But these gestures are part of what makes watching lecture video different than simply looking at the slides. Hence we interpret the laser pointer gestures as we analyze the videos, creating representations that can be transmitted at a low cost. These representations can then be iconified on the client side and displayed clearly.