An end frame tells the video model where to land. When the landing is a clean state, that is the most powerful control you have: the model interpolates from your first still to your last still and fills the motion between them. When the landing is not clean, the same control is destructive, because the model stops simulating and starts sliding pixels toward the target. This lesson is half decision and half template, and the decision comes first because the most common end-frame mistake is making one at all.
You arrive with an approved first frame and its inventory from lesson 6. You leave with either a second still that is a one-pass edit of the first, or a written reason the shot runs first frame only.