【AI教學】Midjourney + Claude AI 輕鬆生成16種MBTI女巫

又一個邊創作邊學英文的機會來了！

上一篇貓娘基努李維獲得很廣大的迴響，許多朋友開始要求說要讓基哥去龍山寺拜拜、還要去傳統市場遛狗，甚至竟然妄想要基哥幫忙清貓砂？

我當然是，絕對不會辜負粉絲們的心意的：

不過玩弄基哥的戲碼就到此為止✋

有認真的朋友偷偷問我說，他看了我的教學後，也想用Midjourney自創角色，但對於形象概念暫時沒什麼想法，而且英文也不太好，很多服飾、髮型單字都不懂，該怎麼辦呢？

這時後就輪到另一個AI Claude出馬啦！由Anthropic開發，一樣是類似ChatGPT的LLM聊天機器人，Claude AI在寫作上又比ChatGPT更加的精準。

很多人抱怨ChatGPT變笨的時候，我都直接轉介大家使用Claude或是Gemini。

Claude角色生成Prompt模板

要生成虛擬角色，我們只要以下prompt模板丟給Claude，就可請它幫忙生成多種不同造型風格的角色敘述了。

先來看看prompt的基礎結構：

a full body shot of a [gender] model [pose], [age], [ethnicity], with [facial features] and [hairstyles], wearing [tops], [bottoms], [shoes], [accessories], carrying [object]

其中 [ ]裡面的分別是人種、年齡、臉部特徵、服飾髮型等，是可以請Claude 幫忙產生的，當然如果你已經有想法，就可以替換成你自己的喜歡的風格造型。

接著將Prompt丟給Claude，請他按照這個結構產生出一個可愛又迷人的反派角色：

Follow this prompt structure and generate a charming antagonist character: a full body shot of a [gender] model [pose], [age], [ethnicity], with [facial features] and [hairstyles], wearing [tops], [bottoms], [shoes], [accessories]

如果已經有了故事腳本或是人物小傳的，也可以補充描述一下角色的個性、興趣喜好等等。提供給Claude資料越詳細，生成的prompt就能越精準。

輸入完成後，Claude就會生給你一段Prompt，描述角色長相穿著等等細節，接著再丟到Midjourney裡面讓它生成角色就好啦：

A full body shot of a model standing, looking directly at the camera, 34 years old, Japanese ethnicity, with almond-shaped eyes, full lips, and a delicate yet haughty expression, with sleek black hair styled in a severe symmetrical bob, wearing a structured white blazer with dramatic winged shoulders over a sheer black lace bodysuit, skintight black leather pants, pointed-toe stiletto knee-high boots, a thick platinum choker and matching cuff bracelet, white background

一次生成多種角色

想要一次生成多種角色也沒問題，例如下面我請他參考12星座，去產出時尚精靈人像：

Follow this prompt structure and create 12 elfs based on 12 zodiac signs: a full body editorial shot of a female elf [pose], 20s, [ethnicity], with long ears and [facial features] and [hairstyles], wearing [tops], [bottoms], [shoes], [accessories], [color] background, in the style of Vogue

或是延伸加入場景和其他配件，請AI幫我做出不同MBTI個性的女巫：

Follow this prompt structure and create 16 characters based on MBTI types: a full body shot of a 20s female witch [pose], [age], [ethnicity], with [facial features] and [hairstyles], wearing [color] witch had and [tops], [bottoms], [shoes], [accessories], carrying [object], in [scene]

好，不過生成快歸快，還是記得仔細看一下Prompt，把該背單字背一背，以後打prompt就不會再手忙腳亂了🤣