2025/4/18 update:这段时间一直在用空闲时间来练习训练SDXL的lora，因为发现以前的Lora在某些情况下存在可能无法还原画师画风的问题，所以我一直都没有把连我都觉得史的lora上传上来。顺便一提，这个lora还是存在一些小问题，不过因为我现在忙于其他事情，所以可能又要消失很久（也许？）权重：1（推荐尝试更高的权重来获得更还原的画风）触发词：防冻液（必须）I’ve been spending my free time lately practicing and training SDXL LoRAs, because I noticed that all of my previous LoRAs sometimes seem to fail to accurately reproduce the artist’s style. That’s why I haven’t uploaded any LoRAs that even I think are terrible.Btw, this LoRA still has some minor issues, but since I’m busy with other things now, I might disappear again for a long time (maybe?).Weight: 1 (recommend try higher weights for a more accurate style reproduction.) Trigger word: 防冻液 (necessary)<hr />本lora的目的是改善noob模型中画师tag的表现，但不代表原模型的画师tag就一定不如lora。训练底膜为<a target="_blank" rel="ugc" href="https://civarchive.com/models/833294?modelVersionId=1116447">noob eps11</a>，仅测试了eps11和vpred10，在eps11表现良好，而在vpred上表现一般。你也可以尝试其他Noob模型或者Illustrious系模型。English:This lora aims to enhance the performance of artist's tag from NoobAI, but it dosen't mean the lora is better than artist's tag.Base model is <a target="_blank" rel="ugc" href="https://civarchive.com/models/833294?modelVersionId=1116447">noob eps11</a>, only tested on eps11 and vpred10. Works great with eps11 and normal with vpred10. You can also try it with other noob models or Illustrious models.

Files with SHA256: 2a6bb8995d4fe80da881ed146b879b4ff5ebb978e84e50471a1ca33751093773