<ul><li>The weight can be used from <code>0.1</code> to <code>1</code></li><li>Sampler: Euler a: 25 to 40 steps.</li><li>CFG Scale: 1 to 7.</li><li>Clip Skip: 2.</li><li>V2减少了训练集图片数量，优化了眼睛（没多大改变），同时个人感觉在黑暗环境下的光线没有V1好(脸部曝光过度）</li></ul>

Files with SHA256: 64bfb1cb4cbd1b334546ac402178a09ca3a78cb441493dbfcd0e87cb0872391c