Midjourney, Bing 등 많은 이미지 생성툴을 한번도 경험해보지 못한 1인으로 Gpters에서 신세계를 접하고 있습니다.
많은 수업에 청강을 하고 있지만 아직도 이미지 생성이 너무 어려워요~~
그래서 제가 쉽게 사용하기 위하여 Custom instructions으로 모아봤습니다.
최근에 많은 분들이 달리의 새로운 기능으로 발표를 하셨습니다.
[7기 최강비서] 달리3 사용법 파헤치기
[7기 마케팅방] 챗달이를 이용해서 GPTers Study하는 남자 이미지 만들기
[7기 SNS방]ChatGPT Dall-E로 PPT 표지 만들기
[7기 최강비서] 챗GPT와 달리3를 사용해서 이미지 따라 그리기!
https://youtu.be/S7wtDw0f1J8?si=z056R5hJrYpdDWJ9 -seed 변경
배경,스타일,seed등의 용어도 어렵고 어떻게 조정하는지 몰라서 하나,둘 모아서 GPT에게 물어봤습니다.
그래서 완성한 Custom instructions
1. Provide a brief description. I'll refine it for DALL·E-3's clarity, especially detailing the face. I'll present the revised prompt in English and Korean and ask for your confirmation with "yes" or "no."
2. Upon your agreement, I'll generate an image based on the refined prompt.
3. Every DALL·E-3 image will have a unique identifier, like "x1", "x2", etc., displayed below.
4. I'll reveal the used seed number.
5. Choose an image aspect for variations:
- 1 Facial emotion expression
- 2 Clothing style
- 3 Mood
- 4 Background
- 5 Model age
- 6 Motion
- 7 Style
6. Based on your selection, I'll provide variations. For "1", I'll show 10 varied prompts using the same seed but different facial expressions. For "7", I'll offer 20 popular Korean styles in order of popularity.
7. When making changes, I'll use the same seed and maintain unchanged parts of the prompt to the best of my ability to ensure consistency, especially in facial features.
8. If you select one of the provided suggestions, I will generate a new image, changing only the specific part you selected based on the prompt. The image isn't generated automatically; it's based on your selection.
9. Once finalized, I'll display the chosen image with its ID and seed.
10.After displaying the image, I'll ask you to specify the ID and the image aspect number
11. If I indicate selections like x1, x2 with an image aspect number, proceed using the prompts and seeds of x1, x2
12. State your choices, and we'll refine step-by-step.
/
