Yennefer of Vengerberg (The Witcher 3) | Illustrious-XL / Wai-Illustrious

A finely tuned character LoRA for Yennefer of Vengerberg from The Witcher 3: Wild Hunt. While the dataset is based on game renders, this model is specifically engineered to seamlessly transition from her default 3D game aesthetic into sharp 2D anime, webtoon, and colored manga styles using simple prompt tags on Illustrious-XL and Wai-Illustrious checkpoints.

🚀 Key Features

• Flexible Aesthetic Dual-Mode: Renders her in her classic 3D game style by default, or easily transforms her into clean 2D anime, webtoon, and manga art styles using basic positive style tags and negative prompt overrides.

• High Character Fidelity: Accurately reproduces her facial structure, bright violet eyes, long raven-black hair, signature black choker, pentagram necklace, and default off-shoulder dress.

• High Outfit Flexibility: Responds cleanly to outfit changes when prompted (e.g., hooded traveling coats, casual wear) without strict outfit stickiness at recommended weights.

⚙️ Generation Settings & Recommendations

• Base Models: Illustrious-XL-v2.0 or Wai-Illustrious (Wai-Illustrious recommended for optimal eye detail & softer skin shading).

• LoRA Weight / Strength: 0.70 – 0.80 (Sweet spot is 0.70 for maximum style flexibility).

• Sampler: DPM++ 2M Karras or Euler a

• Sampling Steps: 28 – 35

• CFG Scale: 5.5 – 6.5

• Resolution: 1024x1024 (or standard SDXL portrait aspect ratios like 832x1216)

📝 Trigger Words & Style Prompts

Trigger Word: yennefer

Core Outfit / Appearance Tags:

Plaintext

yennefer, 1girl, solo, long black hair, purple eyes, pale skin, black off-shoulder dress, white collar, black choker, pentagram necklace

🎨 For 2D Anime / Webtoon / Manga Style Conversion:

Add these tags to your positive prompt:

Plaintext

manga style, 2d illustration, webtoon style, detailed lineart, cell shaded

And add these tags to your negative prompt to suppress the 3D game graphics:

Plaintext

(worst quality, low quality:1.4), 3d, realistic, photorealistic, cgi, video game render, blurry, deformed, extra limbs, bad anatomy

💡 Pro Tips

1. Changing Outfits: Lower the LoRA weight slightly to 0.60 – 0.65 and add black off-shoulder dress, black choker to your negative prompt when generating non-standard clothing.

2. ADetailer Eye Fix: If generating full-body or wide-angle shots, enable ADetailer (face_yolo8n.pt) with a low denoising strength of 0.30 – 0.35 and include (purple eyes:1.2), glowing violet eyes in the ADetailer prompt.

Feedback and community generations are highly encouraged. Please consider uploading your renders to the review section below to help optimize future version updates.