A head mounted display (HMD) identifies viewing trends for a video based on multiple viewings of the video by users having different user characteristics. The analysis identifies viewing trends based on user characteristics, e.g., the age and gender of the user. When a subsequent user is viewing the video, a production & post production module receives the user's profile information and identifies the predicted viewing location of the user at particular times during the video using the identified trends based on where previous users/viewers, who have a profile or characteristics similar to the user, were looking at the particular times during the video. The production & post production module identifies the center tile as the predicted viewing location at a particular time for the user and prepares and encodes the video stream based on the predicted viewing location and, for example, streams the video to the HMD in which the predicted viewing location is the center tile at the associated time in the video. This can increase the effectiveness of the data transmitted to the HUD.