An aspect of the present invention reduces memory accesses while forming combined images of a subpicture frame with each of a sequence of image frames. In an embodiment, line information indicating the specific lines (e.g., rows) of the subpicture frame having display information is first generated. The line information is then used to retrieve only those lines of the subpicture frame that have display information while forming combined frames. According to another aspect, the portions of the combined images having sharp edges due to the inclusion of the subpicture frame are identified based on the line information, and such portions are then filtered. The sequence of combined frames may be displayed on a video output device.