随随便便的融合X-I – v1.0-Inkwash
Run Model | 使用GRAVITI Diffus 关于GRAVITI Diffus |
Model Name | 随随便便的融合X-I |
Model Type | Checkpoint |
Base Model | SD 1.5 |
Version Name | v1.0-Inkwash |
File Name | XI_v10Inkwash.safetensors |
SHA256 | fb5555b9cb |
NSFW | False |
Trigger Words | inkwash style, xianxia style |
Tags | character |
Metadata | fp = fp32 size = full format = SafeTensor |
Size | 4.11 GB |
Version Description |
在硬盘角落里发现早期自训练的Lora,然后随手整合到刚融合的新模型里,诞生出来的一种带水墨仙侠风的风格,看着还行,就分享下了。 machine translation: I found the early self-trained Lora in the corner of the hard disk, and then integrated it into the newly fused new model, and a style with ink fairy style was born. It looked okay, so I shared it. |
Source | https://civitai.com/models/49230 |
Author | WanDou7 |
随手融合的一个模型,比例已经忘记了,只是把之前炼制的lora融合到新模型里:
所用的底模为 revAnimated 1.2.2
增加对比度使用了 Ally’s mix
替换亚洲脸使用了 BRA
烘焙了 clearVae
最后加上自己早期训练的仙侠服装Lora以及水墨风格Lora,
未使用 BWM, 就成了这么一个模型。
可以使用两个触发词:inkwash style (水墨风格) | xianxia style (仙侠风格)。
现在还有点瑕疵,等有空了再调整….自己有基础的,也可以按照上面的逻辑自己融合,比如自己替换其他风格衣服等….
本人郑重声明:
本模型禁止用于训练基于明星、公众人物肖像的风格模型训练,因为这会带来争议,对AI社区的发展造成不良的负面影响。
machine translation:
A model that was fused casually, the proportions have been forgotten, just integrate the previously refined lora into the new model:
The counter used is revAnimated 1.2.2
Increased contrast using Ally’s mix
Replaced Asian face with BRA
baked clearVae
Finally, add Xianxia Lora I trained in the early days and the Ink Wash Lora,
Without using BWM, it becomes such a model.
Two trigger words can be used: inkwash style (ink wash style) | xianxia style (Xianxia style).
There are still some flaws, and I will adjust it when I have time… If you have a foundation, you can also integrate it according to the above logic, such as replacing other styles of clothes by yourself