简介：少女感十足的人物写实Lora模型，此lora是基于majicMIX realisticV7版本为底模进行训练的，对图片数据集的脸部也另外进行了加强训练
Introduction: A realistic Lora model with a strong sense of girlishness. This Lora is trained based on the MajicMIX realistic V7 version as the base model, and has also undergone additional training on the face of the image dataset
关于1.5版本说明：此版本是在1.0版本的基础上少量融合了koreanDollLikeness以修正出图时总倾向于聚焦脸部的问题，算是一个临时迭代版本吧，综合来说是优于1.0版本的，等过段时间不忙了会重新梳理更多的全身照数据集以修正此问题
Regarding version 1.5: This version is based on version 1.0 and incorporates a small amount of koreanDollLikeness to correct the issue of focusing on the face when creating images. It can be considered a temporary iterative version, and overall, it is better than version 1.0. After a while, when I am not busy, I will reorganize more full body photo datasets to correct this issue
召唤词：liqi 这也是我为该lora模型取得名字(璃琪) 不加这个提示词影响也不大的
Summoning word: liqi. This is also why I obtained a name for the Lora model (Liqi). Not adding this prompt word has little impact
lora建议权重推荐 V1.0:（0.5-0.7）V1.5:（0.7-1.0）
Lora suggests weight recommendation V1.0:（0.5-0.7）V1.5:（0.7-1.0）
VAE :vae-ft-mse-84000-ema
底模建议：majicMIX realisticV7版本，墨幽人造人，CNrealistic_MIX_V40其他写实类大模型
chekpoint mold suggestion: MajicMIX Realistic V7 version, Moyou Man made, CNRealistic MIX_V40 other realistic large-scale models
生图参数：DPM++ SDE Karras 迭代步数：35 尺寸大小：512*768
Graph parameters: DPM++SDE Karras Iteration steps: 35 Size: 512 * 768
推荐开启高清修复
Recommend enabling high-definition repair
高清修复设置：放大算法：4X-UltraSharp 重绘幅度 0.3-0.35 高清修复采样 ：30 放大倍数：1.5倍及以上
HD Repair Settings: Amplification Algorithm: 4X UltraSharp Redraw Range: 0.3-0.35 HD Repair Sampling: 30 Magnification: 1.5 times or more
可开启ADetailer脸部优化插件 能有效避免脸崩问题
Enabling the ADetailer facial optimization plugin can effectively prevent face collapse issues
正向起手式提示词推荐：(masterpiece:1.2),best quality,high resolution,(realistic:1.2),(photorealistic:1.4),(solo:1.2),(1girl:1.2),beautiful detailed eyes and face,(detailed skin texture:1),8k,RAW photo 注意：由于脸部加强训练次数较多，在书写提示词时请在起手式提示词的基础上加上一些服饰的描述哦！！！
Recommended positive start up prompt words: (masterpiece: 1.2), best quality, high resolution, (realistic: 1.2), (photosensitive: 1.4), (solo: 1.2), (1Girl: 1.2), beautiful detailed eyes and face, (detailed skin texture: 1), 8k, RAW photo. Note: Due to the high frequency of facial reinforcement training, please add some clothing descriptions to the start up prompt words when writing!!!
第一次将训练出来的模型发布出来心里还真有点忐忑不安，如果有什么不好的地方希望各位大佬予以指正
n I release the trained model for the first time. If there are any negative aspects, I hope everyone can correct them