基于flux Klein 9B训练的支持动漫人物换装的lora模型，可支持多种人物风格，多种服装风格，最后的结果人物比例和服装风格会和原始人物保持一致。
step：4-8，权重：0.5-0.6。人物训练数据是1024：2048，实现效果最好。
prompt：Transfer the outfit from Image 2 to the person in Image 1 on a clean transparent background.Control Image 1 provides the person reference, including the face, hairstyle, identity, body shape, and overall visual style, which must be preserved; Control Image 2 provides the outfit reference, including clothing, headwear, shoes, and accessories. The transferred outfit should match the visual style and overall appearance of Image 1.Do not include props, handheld items, or unrelated extra objects.
输入的图像1是人物图，图像2是服装图
