The model is by my practice lokr, about 10,000 pieces of material, the subsequent fusion of a Pallas's cat model to reduce limb collapse. The effect is not bad. Because the material is mostly human body, so the performance of other aspects is weak. Sampling method: DMP + + 2m Karras or 3M algorithm, iteration steps: 30 to 50 recommended size: 1:1/2:3/3:2/3:4/4:3/9:16/16:9(other sizes are welcome) prompt bootstrap factor (CFG-RRB- : HD0ResolutiontFix fix: 4x-UltraShar

————————————————————————————————————————

p该模型是由我练的lokr，素材一万张左右，后续融合一个兔狲佬的模型，以减弱肢体崩坏。效果还可以。

由于素材大多是人体，所以其他方面表现较弱。采样方法：DMP++2M Karras 或者3m算法

迭代步数：30到50

推荐尺寸：1:1/2:3/3:2/3:4/4:3/9:16/16:9（其他尺寸欢迎尝试）

提示词引导系数（CFG）：7.0

高清分辨率修复：4x-UltraSharp