Translated data: Researchers have introduced a method called FreeControl, which enables spatial control over text-to-image generation models without the need for training. This method supports simultaneous control over multiple conditions, architectures, and checkpoints, achieving control and alignment of generated images through structural and appearance guidance. This innovation is expected to enhance the quality of generation and expand the application areas of AI.