Milana LoRA (WAN 2.1)
Trigger word
Milana
Base model
wavespeed-ai / wan-2.1-14b
Recommended settings
- LoRA weight: 0.7 – 1.0
- Sampler: default
- CFG: 5–7
- Steps: 25–35
Description
Character LoRA trained for consistent appearance and facial features.
Focus on natural expressions, realistic proportions, and clean lighting.
Designed for use with WAN 2.1 base models.
