生成之後，跟它說：「啊我忘記指定圖畫大小了，我要一張橫的」
它會拿提示詞再另外生出一張：

有時候也會和前面那張差蠻多的

A Japanese manga-style illustration of a young woman standing in a clothing store, taking a selfie with her smartphone. She wears casual attire with a light gray tied blouse over a white crop top, dark pants, and white sneakers.

She carries a beige crossbody bag and holds a shopping bag in one hand. The background features store racks with clothing items and mannequins. The character has a calm expression and stands fully visible in a relaxed pose. The illustration emphasizes clean linework, expressive eyes, and manga-style shading. Portrait orientation (1080x1920).

提示詞長這樣，不過有時候又會覺得有點不對，生成圖畫時，好像又有這些提示詞之外的資訊？
比如說整張圖（背景）的色系，提示詞裡好像…沒提到？
但從照片原圖到後來生出來幾張，整張圖的色系是會蠻接近的