I am using a cheap wide angle lens on the iPhone (clip on type), this helps since it is not very high in the air and allows it to "see" more of the ground. I simply just twisted the pole while capturing video.
Once you bring the video into ICE, you can select the start and stop time of the video to do the processing on. The video segment I used was only about 8 seconds long. The application takes all the images in the video (e.g. >15 frames per second) and matches them up and stitches them together to form the larger composite image... magic!