Yae Miko LoRA for SDXL - Genshin Impact Character Generator
Generate Yae Miko from Genshin Impact with this SDXL LoRA using trigger word YeMkAF. Features, settings, and example images.
On this page
Yae Miko LoRA for SDXL
TL;DR
- SDXL LoRA for generating Yae Miko from Genshin Impact
- Trigger word:
YeMkAF - Based on SDXL 1.0, trained on large dataset
- Non-commercial fan creation
What This Model Does
This LoRA (Low-Rank Adaptation) enables Stable Diffusion XL 1.0 to generate consistent images of Yae Miko, the Electro shrine maiden and head of the Yae Publishing House from Genshin Impact. The model was trained on a large dataset with GPU compute sponsored by Redmond.AI.
Trigger Word
Use YeMkAF in your prompt to activate the character. The trigger word is essential for the LoRA to apply Yae Miko's specific features: pink hair, purple eyes, fox ear accessories, and her traditional red and white miko outfit with golden accents.
Recommended Settings
- Base model: SDXL 1.0
- LoRA weight: Start around 0.8-1.0, adjust as needed
- Sampler: DPM++ 2M Karras or Euler a
- Steps: 30-40
- CFG scale: 7-9
- Resolution: 1024x1024 (SDXL native)
The author notes the LoRA "is not perfect and sometimes needs more than one gen to create good images." Multiple generations with the same prompt may yield better results.
Example Prompts
YeMkAF, 1girl, solo, pink hair, purple eyes, miko outfit, fox ears, shrine maiden, high quality, masterpiece, best qualityYeMkAF, Yae Miko, Genshin Impact, standing, torii gate background, sunset lighting, detailed face, intricate clothingLicense & Usage
This is a non-commercial, fan-made creation intended solely for entertainment purposes. All rights to the Yae Miko character belong to HoYoverse (miHoYo). The model author (ArtificialGuyBR) requests that rights holders with concerns contact them at [email protected].
Common Errors and Fixes
Error | Cause | Fix
|-------|-------|-----
Character doesn't resemble Yae Miko | Missing or misspelled trigger word | Ensure YeMkAF is in the prompt Inconsistent results across generations | LoRA stochasticity | Generate multiple images, adjust weight Outfit details incorrect | LoRA imperfection | Try higher LoRA weight (0.9-1.1) or add outfit descriptors
FAQ
Q: What base model does this require? A: SDXL 1.0 (the LoRA was trained on this base).
Q: Can I use this commercially? A: No. This is explicitly a non-commercial, fan-made creation.
Q: Does it work with SD 1.5? A: No, this is an SDXL LoRA and requires an SDXL base model.
Q: Who created this? A: ArtificialGuyBR, with GPU time sponsored by Redmond.AI.
Sources
- Civitai model page - Model card, trigger word, training details, sample images
- Model images from Civitai CDN
This model card was enriched for the ArtificialGuyBR portfolio. Original model by ArtificialGuyBR.