My prompt writing method in a 1,800-character limit environment
JHello? I am a Korean user. I use GPT's help to create prompts in a mix of Korean and English for my work. I tend to pay a lot of attention to details such as poses, lighting, composition, and costumes. I always share my prompts so that others can refer to or use them if they like the images. However, when I upload my work to the gallery, the prompt content is automatically translated into English and displayed.
Recently, I have been expressing various ideas by creating a worldview called Noctrium and a character named Sera who is active in that world. However, while browsing the gallery, I saw an image of Sera posted that I hadn't uploaded myself.
When I checked it, I found that the Noctrium worldview and Sera's settings that I had intended to express were written in the prompt, but the lighting details were not implemented. It seems that they couldn't include the lighting-related content in the prompt due to the 1,800-character limit when generating images.
This site has an 1,800-character limit when generating images, so I often run out of space when I keep adding things I want to include. In such cases, I use a method of mixing Korean and English. Due to its structure, Korean can contain more information in the same number of characters compared to alphabet-based languages. Since it seems that there are the most Japanese users here, I suspect that Japanese might have the same structural characteristics.
Based on my brief experience, I have determined that writing prompts in English is the best way to get intuitive results as desired. However, I think using a language with high information density together can be a viable method in environments with character limits.
I always tell GPT this: write primarily in Korean, but use English only for parts where it is judged that the AI can understand or implement them more accurately.