[Ultra-Practical] The Magical Procedure to Dramatically Improve the Quality of Your "Favorite Cosplay Girl" with AI
Are you enjoying AI image generation?
Have you ever thought, "I want to have my favorite character cosplay!" only to find that "it looks somewhat like the outfit, but the fine details are different" or "it just turned into a girl wearing blue clothes"?
Actually, there is a "little extra step" you can add to make AI perform accurate cosplay, and it will dramatically improve the quality.
This time, I will teach you a search & translation prompt creation method that anyone can easily do!
The secret lies in "image search"!
AI often cannot fully understand the specific design just by using the character name like "[Character Name] cosplay."
That is why it is important for humans to search and verbalize the features. It is a collaborative effort between AI and humans.
The Ultimate Step: Search & Translation Method
* Use Google Image Search or similar to find reference materials (photos or illustrations) of the outfit you want them to wear.
* Pick out the "distinctive parts" of that outfit. (Color, material, shape, accessories, etc. Be as specific as possible!)
* Translate the picked-out words into English and incorporate them into your prompt.
Just by doing this, the way the AI understands your request will be vastly improved.
Practice: Let's Create a Popular Character-Style Cosplay
Now, let's actually use this method to generate a beautiful cosplay girl of a popular game character (this time, Ganyu from "Genshin Impact").
Step 1: Understand the outfit's features through search
As a result of the search, I found that the target outfit has the following features. I will use these as instructions for the AI.
* Horns on the head (horns)
* Black bodysuit (black bodysuit)
* Gold decoration/trim (gold trim)
* Detached sleeves/arm covers (detached sleeves)
* Large bell at the neck (bell choker)
* Blue apron-like skirt (blue panel skirt)
I will put these specific words into the prompt.
Step 2: Creating the prompt code and generating the image
Below is the prompt constructed based on the search results. Copy this as is and try having the AI draw it.
Generation Prompt (Cosplay Spell)
> Masterpiece, best quality, highly detailed, solo, 1girl, beautiful cosplay girl, blue hair, purple eyes, (goat horns on head), wearing complex cosplay costume, (black bodysuit with gold trim), detached sleeves, ornate choker with a large bell, blue apron skirt panel, backless design, fantasy accessories, standing in a mystical fantasy garden at night, soft moonlight, looking at viewer, detailed face.
>
(※Point: By enclosing them in parentheses (), I am emphasizing those elements. I am making it recognize that it is not "just clothes" but a "complex cosplay costume")
Negative Prompt (Excluded Elements)
> lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, ugly, deformed.
>
Generation Result: This is the power of the "Search & Translation Method"!
Here is the image actually generated using the prompt above.
What do you think?
The complex features of the outfit, such as the horns, distinctive sleeves, and gold decorations, which cannot be expressed by just instructing "cosplay," have been brilliantly reproduced.
Furthermore, let's generate a close-up image of this character.
Generated Image 2: Checking details in a close-up
Even in a close-up, the texture of the bell and the details of the hair are clearly depicted. This is the power of the search & translation method!
Summary: Start your own cosplay photoshoot
There is only one point.
Search for the outfit you want them to wear and convey its features in English.
If you can do this, you can create any beautiful cosplay girl you want, from anime characters to movie characters, with your own hands.
Please use this technique to enjoy your own ideal cosplay photoshoot!