V.Beauty girl

• 这是真人训练的模型，使用请遵守当地的法律法规

• 这个模型和浅景深有点拟合

• 这个模特素材全是露牙齿的，素材修复后训练出来的牙齿有问题，下个版本换个模特

• This model is trained by real people. Please use it in compliance with local laws.

• This model fits shallow depth of field to some extent.

• This model's assets are all smiling with teeth showing. The teeth restored and trained from these assets have problems. I will use a different model for the next version.

V幼态美

• 示例图都用了ADetailer

• 触发词：kosina

• 权重什么的自己测试吧，权重1也不会崩

• 还有个露齿笑版，有需要上传，没人要就不传了

V2.0

• 推荐权重 1

• 女孩触发词：xcmt girl , asian girl

• 男人触发词：xcmt , asian man

• v1.0是lycoris，V2.0是lora

• V2.0近照已经达到我想要的效果了，如果还感觉糊用ADetailer 修复，或者放大图片

• V2.0训练尺寸1024*1536

• Recommended weight 1.

• Girl Trigger Words: xcmt girl, asian girl.

• Man Trigger Words: xcmt, asian man.

• V1.0 is lycoris, and V2.0 is lora.

• V2.0's close-up has already achieved the desired effect. If it still appears blurry, use ADetailer for enhancement or enlarge the image.

• V2.0 training size 1024*1536.

V1.0

• 1024*1024素材训练，半身照可以直出，全身照需要 ADetailer 修复

• 质量相关的提示词类似 masterpiece 可以改善脸的效果

• 提示词难调用的话配合插件 LoRA Block Weight 调整te权重使用，以<lora:xcgirl_V4-000002:1:0.5:lbw=FACE>为例，数字1是lora权重，0.5是te权重

• FACE:0,0,0,0,0,1,1,1,1,1,0,0,0,1,1,1,1,1,1,1,0,0,0,0,0,0

• Trained on 1024*1024 data, close up photo can be generated directly, while full-body photos require ADetailer restoration.

• Quality prompts like "masterpiece” is helpful.

• If the prompt words are difficult to work, adjust the TE weight using the LoRA Block Weight plugin. For example, using <lora:xcgirl_V4-000002:1:0.5:lbw=FACE>, where the number 1 represents the LoRA weight and 0.5 represents the TE weight.