Updated: Jun 7, 2024
style
Test 30 multi-model fusion, high degree of integration, reject ugly pictures, abandon character ugliness, resist low pixel and low quality, add skin texture details, strengthen texture performance, and soften the light source! This LoRa belongs to the functional LoRa of differential refining, which mainly works on the skin texture and improves the HD texture. It can be applied to all realistic situations of characters.
Recommended setting parameters: (Newbies can add and modify them based on this)
Basic identification:
• Positive prompt words: Masterpiece level: 1.2), (Best quality: 1.2), (High quality: 1.2), (Realistic: 1.2), (Photo realistic: 1.2), (Official painting: 1.2), (Extremely detailed: 1.2), 1 girl
• Negative cue words: (Worst quality: 2), (Low quality: 2), (normal quality: 2), low definition, watermark, (not suitable for public: 1.2), (Mouth open: 1.6), (teeth visible: 1.6)
Recommended sampling: DPM++ SDE, DPM++ 2M
High resolution fix: Open better