【MBW Warming】

本模型为MBW制作且未使用SuperMerge排列组合修复，会存在包括但不限于：提示词失效、乱加细节、出图解构崩坏、图片模糊等情况。模型的img2img性能与txt2img可能存在差异，使用Hires. fix可能会得到与txt2img直接生成图像不同的结果。

This model is made by MBW and does not use SuperMerge to arrange and combine repair. There will be some problems including but not limited to: prompt word failure, indiscriminate details, breakdown of diagram deconstruction, blurred picture, etc. The img2img performance of the model may differ from that of txt2img. Hires. fix may produce different results than txt2img directly generated images.

————————————————————

本人的所有MBW模型都在这个链接里，不同版本为完全不同的模型而并非升级或者修复，上传只为了方便在多处测试模型性能和能够在别的地方下载。

All of my MBW models are in this link, different versions are completely different models rather than upgrades or fixes, uploaded only for the convenience of testing model performance in multiple places and being able to download elsewhere.

本系列模型不建议继续训练，不建议用于训练LoRA，也不建议使用NAI系列模型所训练的任何LoRA（可能会产生奇怪或者崩掉的输出）

This model is not recommended for continued training, for training LoRA, or for using LoRA trained by NAI series models.(May produce strange or terrible output)

我尝试了一下MBW融合模型，实际使用下来发现这远比Dreambooth等模型训练方式制作模型要费时费力的多。

The model was made for the author trying to start using MBW,In practice, it is much more time-consuming and laborious than model training methods such as Dreambooth.

所有模型都请不要使用任何外置VAE，使用外置VAE会造成包括但不限于生成图模糊、线条出现问题等情况。模型内VAE不存在任何问题，所以也无需使用；出图模糊请先检查是否使用了外置VAE。

VAE does not have any problems in the model, so it is unnecessary to use VAE externally. The use of VAE will cause problems including but not limited to blurring of generated images and line problems.If the outgoing picture is fuzzy, check whether it is external VAE.

所有模型都不建议使用EasyNegative或者badhands之类的emb，更不建议使用Detail Tweaker LoRA，如果在这个模型使用这些东西，将会带来严重的提示词准确度问题或者崩溃的输出。

Embs such as EasyNegative or badhands, much less Detail Weaker LoRA, are not recommended, which, if used in this embs or LoRA, would cause severe hint accuracy problems or terrible output.

KawAICE-Channel这个模型一般意义上非常难以融合，我在ICE版本中尝试修复KawAICE-Channel的对于背景的致命性打击，但是很显然不成功。KawAICE-Channel由于使用特殊的训练方法达到60%以上的画好手的概率，但是这个方面目前无法通过融合的方法融入其他模型中。

The KawAICE-Channel model is generally very difficult to merge. I tried to fix KawAICE-Channel's deadly impact on the background in the ICE version, but it was obviously unsuccessful. KawAICE-Channel achieved a 60% or more chance of completing the "good hand" due to the use of special training methods, but this aspect cannot be incorporated into other models by fusion method at present.

————————————————————

推荐参数 | Recommended parameters:

Anything

• 你可以使用您喜欢的任何采样器、步骤、cfg

  You can use any Sampler, steps, cfg you like

• 比如我喜欢如下参数：

  For example, I like the following argument：

  Sampler: Euler A

  Steps: 20

  CFG: 7

  Clip Skip: 2

• Negatives:You need, not something that's fixed！

  不过为了达到更好的效果，请不要使用EasyNegative

  But for better results, do not use EasyNegative

————————————————————

融合所使用原模型：

HaKoniwalily_cover_style - v1.0 | Checkpoint

苍玄 - 苍玄 | Checkpoint

AICE冰可 | KawAICE[幼态特化模型] - Channel | Checkpoint

万象熔炉 | Anything V5 - V5[Prt-RE] | Checkpoint

Entity_404 - Entity_404 | Checkpoint

SDHK - v2.0 | Checkpoint

T-anime-v2 - v2.0 | Checkpoint

NijiV5 style | Checkpoint（自炼）

Anything V5.2[N] | Checkpoint（自融）

平涂 - Flat coating style | Checkpoint（自炼）

————————————————————

合成配方（YKM）：

[Merge Block Weighted]NijiV5 style+HaKoniwalily_cover_style→test1

Weight_values=1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1

[Merge Block Weighted]test1+苍玄→test2

Weight_values=1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1

[Merge Block Weighted]test2+AICE冰可 | KawAICE[幼态特化模型] - Channel→test3

Weight_values=0,0,0,0,0,0.2,0.2,0.6,0.6,0,0,0,0,0,0.5,0.5,0.2,0.2,0,0,0,0,0,0,0

[Merge Block Weighted]test3+Entity_404 - Entity_404→test4

Weight_values=0,0,0,0,0,0,0,0,0,0,0.5,0.5,0,0.25,0.25,0,0,0,0,0,0,0,0,0,0

[Checkpoint Merger]万象熔炉 | Anything V5 - V5[Prt-RE]+Anything V5.2[N]→Anything V5.2[H]
Weight sum:A*(1-alpha)+B*alpha

alpha=0.67 A=Anything V5 - V5[Prt-RE] B=Anything V5.2[N]

[Checkpoint Merger]Anything V5.2[H]+平涂 - Flat coating style+test4→YKM-beta

sum Twice:(A*(1-alpha)+B*alpha)*(1-beta)+C*beta

alpha=0.8 beta=0.82 A=Anything V5.2[H] B=平涂 - Flat coating style C=test4

[Checkpoint Merger]YKM-beta+NijiV5 style→YKM(U-net)

Weight sum:A*(1-alpha)+B*alpha

alpha=0.1 A=YKM-beta B=NijiV5 style

[toolkit]YKM(U-net)+test4(VAE)+Anything V5.2[N](clip)→YUNOKILLDEME.safetensors

合成配方（YB）：

[Merge Block Weighted]NijiV5 style+HaKoniwalily_cover_style→test1

Weight_values=1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1

[Merge Block Weighted]test1+苍玄→test2

Weight_values=1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1

[Merge Block Weighted]test2+AICE冰可 | KawAICE[幼态特化模型] - Channel→test3

Weight_values=0,0,0,0,0,0.2,0.2,0.6,0.6,0,0,0,0,0,0.5,0.5,0.2,0.2,0,0,0,0,0,0,0

[Merge Block Weighted]test3+Entity_404 - Entity_404→test4

Weight_values=0,0,0,0,0,0,0,0,0,0,0.5,0.5,0,0.25,0.25,0,0,0,0,0,0,0,0,0,0

[Merge Block Weighted]test4+SDHK - v2.0→test5

Weight_values=0,0,0,0,0,0,0,0,0.3,0.3,0.4,0.4,0,0.2,0.2,0.25,0.25,0,0,0,0,0,0,0,0

[Merge Block Weighted]test5+T-anime-v2 - v2.0→test6

Weight_values=0.1,0.1,0.02,0.02,0,0.05,0.05,0,0,0,0,0,0,0.2,0.2,0.25,0.25,0,0,0,0,0,0,0.05,0.05

[Merge Block Weighted]NijiV5 style+苍玄→test7

Weight_values=0,0,1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,0,0

[Merge Block Weighted]test7+Entity_404 - Entity_404→test8

Weight_values=0,0,0,0,0,0.2,0.2,0.6,0.6,0,0,0,0,0,0.5,0.5,0.2,0.2,0,0,0,0,0,0,0

[Merge Block Weighted]test8+anime-v2 - v2.0→test9

Weight_values=0,0,0,0,0,0.2,0.3,0.5,0.1,0.1,0,0,0,0,0,0,0,0,0.6,0.1,0.2,0.2,0.3,0,0

[Merge Block Weighted]test9+平涂 - Flat coating style→test10

Weight_values=0,0,0,0.2,0.2,0.2,0.2,0.2,0.2,0,0,0,0,0,0,0,0,0.2,0.2,0.2,0.2,0.2,0.2,0.2,0

[Checkpoint Merger]test10+test6+万象熔炉 | Anything V5 - V3-RE→YB-alpha
Triple sum:A*(1-alpha-beta)+B*alpha+C*beta

alpha=0.6 alpha=0.35 A=万象熔炉 | Anything V5 - V3-RE B=test6 C=test10

[Checkpoint Merger]万象熔炉 | Anything V5 - V5[Prt-RE]+Anything V5.2[N]→Anything V5.2[H]
Weight sum:A*(1-alpha)+B*alpha

alpha=0.67 A=Anything V5 - V5[Prt-RE] B=Anything V5.2[N

[Checkpoint Merger]Anything V5.2[H]+平涂 - Flat coating style+YB-alpha→YB-beta

sum Twice:(A*(1-alpha)+B*alpha)*(1-beta)+C*beta

alpha=0.8 beta=0.82 A=Anything V5.2[H] B=平涂 - Flat coating style C=YB-alpha

[Checkpoint Merger]YB-beta+NijiV5 style→YB(U-net)

Weight sum:A*(1-alpha)+B*alpha

alpha=0.1 A=YB-beta B=NijiV5 style

[toolkit]YB(U-net)+test6(VAE)+Anything V5.2[N](clip)→YUNOBEST.safetensors

合成配方（ICE）：

[Merge Block Weighted]YB+AICE冰可 | KawAICE[幼态特化模型] - Channel→ICE（U-net）

Weight_values=？

[toolkit]ICE(U-net)+YB(VAE)+AICE冰可 | KawAICE[幼态特化模型](clip)→YUNOICE.safetensors

————————————————————

模型可以和其他模型一样随意使用，但是请遵守所在地区的法律法规，以免造成麻烦（我们并不负责）

The model can be used as freely as other models, but please abide by the local laws and regulations, so as not to cause trouble (we are not responsible for it)