"Kamado Tanjirou" from "Demon Slayer" Lora model.
Model Introduction（模型介绍）
Hello everyone! I'm excited to announce the release of my Lora model on C Station for the first time. Overall, compatibility seems good after comprehensive testing. It's capable of producing impressive results with various commonly used large-scale models. Feeling confident in its performance, I've decided to share it. Running it with my parameters should yield satisfactory results, and there are still many weights to be adjusted for further testing. If you like it, feel free to experiment with tweaking Lora's weights and share your feedback in the forum to help me improve!
哈喽大家好！第一次在C站发布自己的Lora模型，整体测试下来兼容性不错，多种常用的
大模型都能跑出不错的效果，感觉还可以就发出来了，按照我的参数跑效果不会差，还有很多的权重没有拉高去测，大家喜欢可以尝试着适当调整Lora权重也发反图区让我开开眼!!!!
All images showcasing my Lora model are unedited, without any control net or Photoshop processing. They're pure original images. Before running any images, make sure to check out my model introduction thoroughly! If you're still not getting good results, feel free to reach out to me via email or QQ.
我的lora模型图片展示都是未添加controlnet或者p图处理的,纯纯原图,
跑图之前一定先看我的模型介绍！！！！！！还是跑不出好的效果可以通过邮件或qq联系我,
Sampler采样器：dpm++2m karras
steps步数：25~40看着来吧主要是没往高的测，emmmmm.....
CFG Scale提示词引导：8
image dimensions推荐尺寸：512x768/1：1
Hires.fix高清修复：R-ESRGAN 4x+ Anime6B 0.3~0.4
ADetailer:face_yolov8s.pt
Trigger Words触发词：tzl
The checkpoint model that complements Lora the best: "end" is available for direct download on C Station! It's the checkpoint model with the highest fidelity to the original artwork, as per my testing!
和Lora契合的大模型：《end》c站可以直接下哦，宝藏大模型！！在我测试下来和原画还原度最高的大模型！
Lora's weight: Recommended range is 0.6 to 0.8, but this is just a rough estimate. You should adjust the weight based on the checkpoint model you're using and the desired outcome. If you just want a good outcome,, you can use the recommended weight of 0.8.
Lora权重：推荐0.6~0.8，其实这个只是一个大概的范围，具体要根据自己使用的checkpoint模型和自己想要的效果去调整权重，如果懒的话可以用我推荐的模型0.8去跑,
some useful prompt一些好用的提示词：
'fighting stance,angry,injury,flaming Knife,bleed,Wield a knife,Breath of Fire,Fire magic,Water jet knife,Water magic'.
These words can be combined to create some cool fight scenes. Everyone can try experimenting with different combinations, or they can refer to the hint words in the pictures I've posted.
这几个提示词可以有一些比较炫酷的打斗画面出现，怎么组合大家可以去尝试尝试，也可以参考我发的图片的提示词,
A few minor regrets: The sword's effect isn't as good as I'd hoped, there are always some small imperfections. I'll work on perfecting it when I have time later on.
一些小遗憾：刀的效果不是很很好总有一些小瑕疵，后期有时间我再去完善,
Thank you all for your support! Thank you to Civitai platform for the support as well!
感谢大家的喜欢！！感谢civitai平台的支持！！
描述:
Model Introduction（模型介绍）
Hello everyone! I'm excited to announce the release of my Lora model on C Station for the first time. Overall, compatibility seems good after comprehensive testing. It's capable of producing impressive results with various commonly used large-scale models. Feeling confident in its performance, I've decided to share it. Running it with my parameters should yield satisfactory results, and there are still many weights to be adjusted for further testing. If you like it, feel free to experiment with tweaking Lora's weights and share your feedback in the forum to help me improve!
哈喽大家好！第一次在C站发布自己的Lora模型，整体测试下来兼容性不错，多种常用的
大模型都能跑出不错的效果，感觉还可以就发出来了，按照我的参数跑效果不会差，还有很多的权重没有拉高去测，大家喜欢可以尝试着适当调整Lora权重也发反图区让我开开眼!!!!
All images showcasing my Lora model are unedited, without any Controlnet or Photoshop processing. They're pure original images. Before running any images, make sure to check out my model introduction thoroughly! If you're still not getting good results, feel free to reach out to me via email or QQ.
我的lora模型图片展示都是未添加controlnet或者p图处理的。纯纯原图。
跑图之前一定先看我的模型介绍！！！！！！还是跑不出好的效果可以通过邮件或qq联系我。
Sampler采样器：dpm++2m karras
steps步数：25~40看着来吧主要是没往高的测，emmmmm.....
CFG Scale提示词引导：8
image dimensions推荐尺寸：512x768/1：1
Hires.fix高清修复：R-ESRGAN 4x+ Anime6B 0.3~0.4
ADetailer: face_yolov8s.pt
Trigger Words触发词：tzl
The checkpoint model that complements Lora the best: "end" is available for direct download on C Station! It's the checkpoint model with the highest fidelity to the original artwork, as per my testing!
和Lora契合的大模型：《end》c站可以直接下哦，宝藏大模型！！在我测试下来和原画还原度最高的大模型！
Lora's weight: Recommended range is 0.6 to 0.8, but this is just a rough estimate. You should adjust the weight based on the checkpoint model you're using and the desired outcome. If you just want a good outcome,, you can use the recommended weight of 0.8.
Lora权重：推荐0.6~0.8，其实这个只是一个大概的范围，具体要根据自己使用的checkpoint模型和自己想要的效果去调整权重，如果懒的话可以用我推荐的模型0.8去跑。
some useful prompt一些好用的提示词：
'fighting stance,angry,injury,flaming Knife,bleed,Wield a knife,Breath of Fire,Fire magic,Water jet knife,Water magic'.
These words can be combined to create some cool fight scenes. Everyone can try experimenting with different combinations, or they can refer to the hint words in the pictures I've posted.
这几个提示词可以有一些比较炫酷的打斗画面出现，怎么组合大家可以去尝试尝试，也可以参考我发的图片的提示词。
A few minor regrets: The sword's effect isn't as good as I'd hoped, there are always some small imperfections. I'll work on perfecting it when I have time later on.
一些小遗憾：刀的效果不是很很好总有一些小瑕疵，后期有时间我再去完善。
Thank you all for your support! Thank you to Civitai platform for the support as well!
感谢大家的喜欢！！感谢civitai平台的支持！！
训练词语: tzl
名称: tzlNICE.safetensors
大小 (KB): 73847
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
