20 images
Introduction The "Xu'er - Red Dust White Dream" (【无需LORA画主角】绪儿-红尘白梦) is a powerful SDXL checkpoint model designed for generating high-quality anime and game characters directly, without the need for additional LORA models. This V3 iteration significantly improves character generation, particularly addressing common issues with hands and weapon handling. Key features include:
- Direct Character Generation: Create detailed anime and game characters straight out of the box.
- Enhanced Anatomy: Substantially improved hand rendering and more stable depiction of characters holding weapons.
- Dynamic Poses: Capable of generating scabbards and sword-drawing actions, allowing for more dynamic character poses.
- Broad Character Support: Optimized for a wide range of anime and game characters, including those from popular series like Genshin Impact, One-Punch Man, Bleach, Code Geass, Nier: Automata, Steins;Gate, Dragon Ball, Re:Zero, Attack on Titan, Sailor Moon, Berserk, Spy x Family, Naruto, One Piece, Demon Slayer, Overlord, Fate/stay night, JoJo's Bizarre Adventure, Neon Genesis Evangelion, Uma Musume Pretty Derby, and Azur Lane. This model excels at producing vibrant, detailed character art with a distinct anime and game aesthetic.
How to Use * Trigger Word: No specific trigger word is required to activate the model's style.
- Recommended Settings: * Steps: 20 * CFG Scale: 7 * Model Weight: 0.8 * Initial Resolution: 850x1280 (for portrait images) * Sampler: Flexible. 2M series samplers offer moderate results, while Euler generally produces cleaner outputs. * Denoising Strength (for img2img/inpainting): 0.4-0.5. This can often help resolve minor hand issues during refinement.
- Tips for Character Generation: * To generate specific anime or game characters, use the format:
Character Name (Anime Name). For example,shenhe (genshin impact). * This model is designed for direct output; all provided example images are generated without LORA or ControlNet. * While LORAs designed for SDXL and PONY models may be compatible, their effectiveness with this checkpoint might be reduced by approximately 80%.
Example Prompts To generate a specific character, use the format character_name (anime_name) within your prompt. Example:
shenhe (genshin impact), 1girl, white hair, long braid, traditional chinese clothes, elegant, detailed face, standing, outdoor, mountains, clear sky, highly detailed, masterpiece
FAQ
Q: What trigger word should I use?
A: No trigger word is required to activate the "Xu'er - Red Dust White Dream" model.
Q: What are the recommended settings?
A: The recommended settings are: Steps: 20, CFG Scale: 7, and Model Weight: 0.8.
Q: What resolution works best?
A: The recommended initial resolution for portrait images is 850x1280.