Memory Management
Provided with a very large input image (for example, 16K x 16K “image moscaic” from an astronomy sky survey) we want to produce a video that captures the movement of a much smaller window across this large image. You can think of this as lens that allows us to focus on a trajectory of interest. To do this, we need to position a rectangular window within this large image and extract all the pixels that overlap it to produce one frame. As we shift the window to a new offset position within the large input image, we will produce a sequence of image frames that can be written out and passed into a video creation tool such as FFMPEG to create a video.