Based on the input image, it generates an image of the SD3.5 Large model accelerated by TensorArt-TurboX, which is automatically tagged by MiaoshouAI tagger, so it works without prompting. Images of people tend to go wrong, so do not expect too much.