A complex Chinese prompt (quoted in the tweet) telling the model to turn the person from image 1 into a real human holding a figure from image 2, wearing clothes from image 3, makeup and hair from image 4, expression from image 5, and with a car from image 6, used for advanced multi-reference testing.
把图一人物1:1变成真人,手里拿着图二的手办,穿着图三人物的穿搭,化着图四人物的妆容和发色,表情是图五的表情,背后停着辆图六的车

