Explanation
For short-form the useful framing is that it removes the floor price of a shot. Filming has a fixed cost per setup regardless of how much the resulting clip is worth, which is why most brands only ever produce footage for the handful of products, angles or messages that clear that floor. Generation makes the twentieth variant cost the same as the first, which changes what is worth trying rather than just what is cheaper.
Control is what separates usable output from a novelty. Being able to pin the first and last frame, hold a product or a face consistent across clips, and choose a shot length is the difference between footage you can cut into a reel and footage you can only post as a curiosity.
Length is a real constraint. Most models produce a few seconds at a time, so a longer video is assembled from shots rather than generated in one pass, which is closer to how editing already worked.
The common mistake
Writing one very long prompt and expecting a finished video. The reliable method is short shots, generated separately, with the frames you care about pinned.