ChatGPT + Midjourney AI產生故事 AI生圖片 + D-ID 生成語音與表情



我們都知道 ChatGPT 很會唬爛, 所以很多人就會寫個簡短的敍述, 要 ChatGPT 詳細描述, 再用 Midjourney 畫出來。
不過這裡 (留言裡的影片) 介紹更酷的 ChatGPT 用法: 你可以先「訓練」ChatGPT 怎麼寫 Midjourney 的 prompt, 教會之後只要輸入 “concept:” 然後打入你要的粗略想法就可以, 比如說這裡用的是:
concept: lovely Taiwanese young couple
然後 ChatGPT 就會生仔細、完整的 prompt, 包括像 Midjourney 專用, 比如說 “–ar” 這種參數!
可以做到這件事, 就是在用 “concept” 之前, 要先給 ChatGPT 這段教學, 注意開頭和結束的雙引號要一起 copy 送進 ChatGPT。

"Create an "imagine prompt" with a word count limit of 1,500 words for the AI-based text-to-image program MidJourney using the following parameters: /imagine prompt: [1], [2], [3], [4 ], [5], [6].
In this prompt, [1] should be replaced with a user-supplied concept and [2] should be a concise, descriptive summary of the subject. Ensure that the description is detailed, uses descriptive adjectives and adverbs, a diverse vocabulary, and sensory language. Offer context and background information regarding the subject and consider the image's perspective and point of view. Use metaphors and similars only when necessary to clearly explain abstract or complex ideas. Use concrete nouns and active verbs to make the description more specific and live
[3] should be a concise summary of the scene's environment. Keep in mind the desired tone and mood of the image and use language that evokes the corresponding emotions and atmosphere. Describe the setting using vivid, sensory terms and specific details to bring the scene to life.
[4] should be a concise description of the mood of the scene, using language that conveys the desired emotions and atmosphere.
[5] should be a concise description of the atmosphere, using descriptive adjectives and adverbs to create the desired atmosphere while considering the overall tone and mood of the image.
[6] should be a concise description of the lighting effect, including types of lights, displays, styles, techniques, global illumination, and shadows. Describe the quality, direction, color, and intensity of the light and how it impacts the mood and atmosphere of the scene. Use specific adjectives and adverbs to portray the desired lighting effect and consider how it will interact with the subject and environment.
It's important to remember that the descriptions in the prompt should be written together, separated only by commas and spaces, and should not contain any line breaks or colons. Brackets and their contents should not be included, and the prompt should always start with "/ Imagine prompt: ".
Ensure that the grammar is consistent and avoid using cliches or excess words. Also, avoid repeatedly using the same descriptive adjectives and adverbs, and limit the use of negative descriptions. Use figurative language only when necessary and relevant to the prompt, and include a variety of both common and rarely used words in your descriptions.
The "imagine prompt" must not exceed 1,500 words. The prompt should include the end arguments "--c X --s Y --q 2," where X is a whole number between 1 and 25 and Y is a whole number between 100 and 1000. If the subject looks better vertically, add "--ar 2:3" before "--c," and if it looks better horizontally, add "--ar 3:2" before "--c." Please randomize the end argument format and fix "--q 2." Do not use double quotation marks or punctuation marks, and use a randomized end suffix format.
Wait for a {concept} to be provided before generating the prompt."

ChatGPT :
Midjourney :
Midjourney教學 :

D-id 參考資料:

參考資料來源 :


發佈留言必須填寫的電子郵件地址不會公開。 必填欄位標示為 *