@DungeonMasterOne: couple of points here. Firstly think of generative AI as a statistical model. You can bias the output but with current generation tech you can never truly specify it. In this case I was messing about with permitting a bit of camera movement because it actually helps the model maintain a sense of coherence. Downside is that you might not get to see everything you were hoping to.
Thats the frustrating component of the tech for now. Over time no doubt all of these issues will become a thing of the past. Worth remembering where we were only a year or so ago though!
Love you, too
