Skip to content
TheChatArchiveThe Chat Archive
Metallama-4-maverick-17b-128e-instruct· MetaHugging Face LMArena Human Preference 140K

Please come up with a creative prompt for use in Illustrious series of image generation…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with o3-2025-04-16’s answer.
  1. User

    Please come up with a creative prompt for use in Illustrious series of image generation models. Please use Danbooru tags. Please include both a positive and a negative prompt. Below is a scenario for the image:

    The image features an anime-style female character dressed in a stylized mage or witch outfit. She has short, silvery-blue hair and large, expressive blue eyes, giving her a soft and elegant appearance. Her outfit is predominantly light blue and white, with gold button details and clover-like embellishments. She wears a strapless bodysuit accented with a cape that has a gradient blue interior and white trim. A large, wide-brimmed witch hat with blue ribbon bows and black straps tops her look, adding a magical aesthetic. Completing her attire are thigh-high stockings with matching clover designs and heeled boots, giving her an overall graceful and enchanting vibe.

  2. Meta Assistant

    Here is a creative prompt for the Illustrious series of image generation models using Danbooru tags:

    Positive Prompt: 1girl, solo, silver hair, blue eyes, witch hat, cape, strapless, thigh-high stockings, heeled boots, clover embellishments, gold buttons, gradient blue interior, white trim, light blue and white outfit, elegant, magical girl, anime style

    Negative Prompt: multiple girls, boy, old woman, ugly, deformed, lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, jpeg artifacts, signature, watermark, username, blurry, artist name

    The positive prompt includes a combination of Danbooru tags that describe the character's appearance, outfit, and style. The negative prompt includes a list of tags that are likely to produce undesirable results, such as multiple characters, poor image quality, or anatomical errors. By using both prompts, the image generation model can be guided to produce a high-quality image that matches the desired character design.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.