This lets you apply the Wai anime style (from my WaiSnackMix checkpoint) to your generations, ensuring a closer match to that target style. The beloved WaiSnackMix checkpoint, founded on the most pleasing balance of color and vibrancy, with detailed anime faces and hair, now comes in a LoRA.This model was trained on a variety of subject matter. What it generates is a style, the subject matter is mostly affected by the way the prompt tokens are interpreted by the model at inference.<h3 id="moss-generation-guide">Moss Generation Guide</h3>Recommended Strength: 0.9-1.2CFG: 3.0-4.0Steps: 30-50Use Moss base model or SFT, merges tend to degrade quality or do not take the styleKeyword: wsm_styleSampler/Scheduler: I recommend Euler Ancestral + Simple or SGM Uniform)<h3 id="anima-generation-guide">Anima Generation Guide</h3>Recommended Strength: 1.0Keyword: wsm_style Other quality tags are sure to have an effect and could cause drift toward or away from your target style.This LoRA is only intended for Wai style, I don't know why you would use it while trying to do realism. Untested.

Files with SHA256: 5d6d2f98a71d22ab7c3e35a69aa2c6e20285b04f76858f8a627f6c4029412b7b