All characters in every scene will appear in full-body front view; JSON will contain maximum detail for all of them and the image prompt will always be wide, head-to-toe, centered frame. Scene objects or new features can also be added to JSON. Character/clothing/environment continuity must be maintained.