Core Mechanics
Danbooru Tagging System
Unlike Midjourney which prefers natural language sentences, our Anime and stylized Uncensored models respond best to comma-separated keywords known as "Danbooru Tags".
Pro Tip: Weighting
You can emphasize or de-emphasize a tag using parentheses and numbers. For example: (red eyes:1.4) forces the AI to prioritize red eyes, while (scarf:0.6) makes it less likely.
The Perfect Formula
1. Quality Tags (Always First)
masterpiece, best quality, ultra-detailed, highres, 8k resolution, cinematic lighting,
2. Subject & Physical Traits
1girl, solo, tall female, athletic body, long purple hair, messy hair, glowing yellow eyes, (intricate cybernetic arm:1.2),
3. Attire & Pose
wearing tactical gear, techwear jacket open, looking at viewer, over shoulder pose, dynamic stance,
4. Environment & Context
raining, tokyo street background, neon signs, depth of field,
Final Assembled Prompt
masterpiece, best quality, ultra-detailed, highres, 8k resolution, cinematic lighting, 1girl, solo, tall female, athletic body, long purple hair, messy hair, glowing yellow eyes, (intricate cybernetic arm:1.2), wearing tactical gear, techwear jacket open, looking at viewer, over shoulder pose, dynamic stance, raining, tokyo street background, neon signs, depth of field