ChatGPT 給出的 MidJourney 提示詞,精準的原因竟然是…
一、前言:
相信透過石頭哥這三個月密集的分享,你一定也對於Midjourney 文字生圖的魔力產生了無比的好奇吧!但是腦袋瓜子能擠出來的《咒語》,總是三三兩兩不成文,相當困擾你嗎?可是召換出chatGPT幫你一把時,總是沒辦法讓ChatGPT精準給出《咒語》?
來吧,今天石頭哥為你揭開精準的面紗吧!看看石頭哥的精準秘密是什麼?同時也帶你看看透過ChatGPT生成的《咒語》,讓你見識一下原來精準的原因竟然是….。
二、召換神奇的ChatGPT,幫忙激盪你的思緒吧!
準備好你的創造力了嗎?在我們開始召換神奇的ChatGPT之前,石頭哥讓你看看,我嘗試三個多月學習理解的精準是什麼?答案是:結構化。
好比我這次給ChatGPT的兩個案例,都是透過結構化呈現,每一個案例都包含、類型、情感、場警、類型、鏡頭、特效、標籤。當你習慣如此下達《咒語》給Midjourney的話,同樣的你提供給ChatGPT參考,它才會依據你的模式提供出更精準的《咒語》,你不妨試試看吧。
IMAGE_TYPE:微距特寫 |類型: 時尚 |情感:迷人 |場景:一個女孩坐在椅子上,穿著時尚的素色五分袖上衣,搭配印花花朵圖案的連衣裙,看起來相當簡約時尚 |演員: 女孩 |位置類型:特寫 |相機型號: 佳能 ESO R3 |佳能鏡頭:100mm f/2.8微距 |特效:景深 |標籤: 微距, 女孩, 景深,五分袖上衣,印花花朵圖案的連衣裙,流行趨勢,簡約時尚

✅Step 1. 先餵食 ChatGPT 兩張圖示的《咒語》:
🍁IMAGE_TYPE: Macro close-up |Genre: lookbook, kpop, korean style |Emotion: Charming |Scene: A girl sits on a chair, wearing a stylish plain five-point sleeved blouse, with a printed one-piece dress with a printed pattern of flowers, which looks quite simple and stylish, which is a popular trend this spring |Cast: girl |Location Type: Close Up |Camera Model: Canon ESO R3 |Canon Lens: 100mm f/2.8 Macro |Special Effects: depth of field |Tags: Macro, girl, depth of field, long-sleeved lace shirt, high-waisted showerhead skirt, trend, simple fashion

🍁IMAGE_TYPE: Macro close-up |Genre: lookbook, kpop, korean style |Emotion: Charming |Scene: A girl leaning against the railing of the fence, wearing a fashionable plain long-sleeved lace shirt and a high-waisted showerhead skirt, looks quite simple and stylish, which is a popular trend this spring |Cast: girl |Location Type: Close Up |Camera Model: Canon ESO R3 |Canon Lens: 100mm f/2.8 Macro |Special Effects: depth of field |Tags: Macro, girl, depth of field, long-sleeved lace shirt, high-waisted showerhead skirt, trend, simple fashion
IMAGE_TYPE:微距特寫 |類型: 時尚 |情感:迷人 |場景:一個靠在欄杆上的女孩,穿著時髦的素色長袖蕾絲襯衫和高腰蓮蓬頭裙,看起來相當簡約時尚,這是今年春天的流行趨勢 |演員: 女孩 |位置類型:特寫 |相機型號: 佳能 ESO R3 |佳能鏡頭:100mm f/2.8微距 |特效:景深 |標籤: 微距, 女孩, 景深,長袖蕾絲襯衫,高腰蓮蓬頭裙,流行趨勢,簡約時尚
總算讓你看到石頭哥以往的《咒語》,竟然是有系統化、結構化的方式展現了吧,相信對於如何下達《咒語》困擾的人,今後應該頗有幫助的,也期望看到你創建令人瞠目結舌的好作品哦!
✅Step 2. 請它參考上述兩個提示,隨機產生三個提示:
create 3 prompts with random parameters by using the same construct as 2 prompts below: ( 使用與以下 2 個提示相同的構造,使用隨機參數建立 3 個提示 )
很棒,chatGPT產出的三組圖片,都相當的令我滿意,首先先看看第一組:相當令人驚豔的黑白照片,這是我以往較少觸及的創作方式,以後確實可以參考一下。
IMAGE_TYPE: Portrait | Genre: Fashion | Emotion: Moody | Scene: A woman standing in a dimly lit alleyway, wearing a black leather jacket, boots, and a black and white outfit, with her messy hair falling over her face. Her expression is serious and contemplative. | Cast: Woman | Location Type: Outdoor | Camera Model: Nikon Z6 | Camera Lens: 50mm f/1.8 | Special Effects: Black and White | Tags: Moody, Fashion, Black and White, Leather Jacket, Boots, Outdoor, Street Photography

第二組圖片呈現開朗燦爛笑容的女孩,還算不錯。
IMAGE_TYPE: Candid | Genre: Documentary | Emotion: Happy | Scene: A woman walking through a busy street, wearing casual summer clothes, a backpack, and her hair in a messy bun. She has a big smile on her face as she looks around at the bustling city. | Cast: Woman | Location Type: Outdoor | Camera Model: Fujifilm X-T4 | Camera Lens: 35mm f/1.4 | Special Effects: None | Tags: Happy, Documentary, Candid, Summer Clothes, Backpack, Street Photography

第三組,這組照片相當令我滿意,我想以後我也會嘗試使用它的提示詞來創作。
IMAGE_TYPE: Editorial | Genre: High Fashion | Emotion: Serious | Scene: A woman standing in a dimly lit room, wearing an elegant evening dress and high heels. Her hair is pulled back into a sleek ponytail, and she has a serious expression on her face. The lighting is dramatic and highlights the curves of her dress. | Cast: Woman | Location Type: Indoor | Camera Model: Sony A7S III | Camera Lens: 85mm f/1.4 | Special Effects: Dramatic Lighting | Tags: Editorial, High Fashion, Serious, Evening Dress, High Heels, Indoor

三、小結:
看完石頭哥結構式的提示詞,是否覺得收穫不少了呢?同時又吸收了如何下達指令給ChatGPT,讓ChatGPT精準理解你要的是什麼相當的重要,如此才能幫助你腦袋激盪出更多的創意點子。看完,這五組石頭哥的《咒語》,是否覺得此篇文章讓你享受到一魚雙吃的喜悅呢?來吧,讓我們因分享而快樂,因快樂而更樂於學習,一起愉快的在AI創作的道路上奔馳吧。
圖片與文章來自:(https://www.potatomedia.co/post/ce1c4dc7-e1aa-47c1-b165-dffab542e973)