v2 (MOSS-Image) 2026.10.21

Hi there, after another so long long time, I finally, and successfully trained another version of this LoRA based on the MOSS-Image-Base!

This time, you do not need any another trigger words, however, all the clothing, hair style, should write yourself into natural language is ok.

我又回来了，这次先用MOSS-Image-Base做一个v2的另一个实验性LoRA，因为时间紧所以打标偷懒了，好处是无需任何触发词，不足就是任何细节和特征描述都需要你来写（已经准备好v2.1重制优化版了，现在还在调试我自己本地的新的打标系统x），建议参照我发送的示例图片来写，一段英文自然语言足矣。总之还是感谢大家支持！:D

After so long long time of being busy on my university studying, I finally have a short vacation to train and then post my LoRAs Again! Being sorry here to let u wait so long :(
This time I select Anima Base v1.0 as the DiT so that my laptop could have much less pressure on VRAM.
The core trigger word is: nakamachi arale. And due to some reasons (I am careless...), this time u should STILL USE DanBooru-Like tags as your prompt ONLY.
And the strength of the LoRA should be 0.66~1 (? I am not sure, so you should try and get the best results by yourself. )

哈哈嗨，时隔不知道多长时间，终于又有空发新LoRA了，这次带来的是梦限大的主唱阿拉蕾
（喜报，这就是梦限大.jpg）
核心触发词就是nakamachi arale，加载强度比我想象的要小很多（0.66~1？自己试试），当然了，由于我没有考虑全面，所以还是跟之前一样，只用tags，用自然语言会多多少少有崩坏的地方（？）
总之就是让大家久等了！以后我如果有空还会继续炼丹！阿里嘎多~🎶 :D