In a recent conversation I had with a KYC implementor for mobile it came out that these "fit your head into the frame / make a gesture" prompts are indeed to get unfakable images of the user, but the real use is in the attempt to fulfil the task. The app is actually recording video and not just trying to get an image under arbitrary conditions. It was explained to me that the full process of adjusting your phone + position within a frame is much harder to fake convincingly than a static image. Your reaction to the prompts, rather than your fulfilment, is what is valuable.