If you use software to position objects in a the frame
And if you simply describe the position via a prompt (among other things) then is that not also minimally creative?
What matters is the human’s input. You can create uncopyrightable works using any tool, and you can use any tool to create copyrightable works. What matters is how much human expression is involved.
Yes, that has basically been my argument - the human input has to be creative in some way and IMO and prompt can be. But not all prompts used will meet that bar. Where the line lies on what meets that bar is still up for debate and AFAIK no court has laid any groupd work for this yet. But a prompt alone can IMO contain enough creativity to allow the AI generated work to be copyrightable.