1 is the large model automatically generating prompts, 2 is using self-written prompts. Select 1 to not use prompts. When choosing 1, you can add elements of the prompts you hope to generate. There will be two results: one before upscaling and one after upscaling. True is pose reference, false is depth reference. The pose is not exactly the same, it is just a reference. The face will be completely replaced, not referenced.