Open Model Radar · Dossier
A7
Character/people LoRA by Sylvora, built on the Krea 2 Turbo base model, with no fixed trigger words.
LoRAKrea 2
Highlights
- No trigger words needed; a plain description is enough
- 14 total downloads · 1 like · 1 version
- File size about 435.9 MB
- Platform-tagged as general content
Scope
- Best for: character/people subjects with the Krea 2 base model
- Best for: stacking with other LoRAs on the same base model (lower each weight by 0.1–0.2)
- Not suitable for: direct cross-base-model use — feeding a Krea 2 Turbo LoRA to another base model degrades output noticeably
Tips
- Start at weight 0.8; higher values overfit
- If output looks washed out or overly stylized, lower the weight first rather than switching samplers
- The base model must be Krea 2; choosing the wrong base model is the most common cause of failed results
Facts
- Type
- LoRA
- Base model
- Krea 2
- Org
- Sylvora
- Downloads
- 14