V2:清洗了数据集，采样更偏低噪训练。Cleaned up the training dataset and adjusted the timestep sampling to favor lower-noise training.__________________________________________________________________________________<h1 id="recommended-settings">推荐参数/Recommended Settings</h1><h3 id="for-anima:">FOR ANIMA:</h3><ul><li>Trigger word:@ezk28</li><li>LoRA weight: 0.8~1</li><li>Sampler: er_sde + beta</li></ul><h2 id="tips-for-v1:">tips for v1:</h2>推荐正面提示词/Recommended Positive Prompts： uncensored, subtle film grain, soft focus, slight chromatic aberration, digital noise, low saturation, muted colors, burry,

Files with SHA256: 640e8e2cdbf2972bf9b908f6316640efa497edc9f07a1dc2a77eef07ceed50a8