Leaves and vines（树叶藤蔓衣）
Details
Download Files
About this version
Model description
欢迎使用 这个树叶藤蔓效果服装模型 此模型兼容 真人，2d，2.5d绘画风格的大模型模型和lora模型
Welcome to this leaf vine effect costume model which is compatible with live, 2d, 2.5d painting style large model models and lora models
-----------------------------------------------------------------------------------------------------
建议出图参数(Suggest drawing parameters):
推荐SDXL底模(Recommended SDXL base mode:)：
Kohaku-XL
StarlightXLAnimated_v3
ArienMixXL Asian portrait
LORA 建议权重0.6~0.8 LORA recommends a weight of 0.6 to 0.8
采样方法（Sampler）:
DPM++ 3M SDE Exponential 45步
提示词引导系数 (CFG Scale): 3~7
人像人物分辨率直出(Portrait character resolution)： 768x1280 ，1024x1024
可以直出 配置够的情况下建议开高清修复
it is recommended to open HD repair if the direct configuration is enough
高清修复（Hires.fix）: 4x-UltraSharp , 高分迭代步数 1530 重绘幅度 0.3 0.5
Clip Skip: 2
(Negative)负面：text,username,logo,(low quality, worst quality:1.4),(bad anatomy),(inaccurate limb:1.2),bad composition,inaccurate eyes,extra digit,fewer digits,(extra arms:1.2),watermark,multiple moles,mole on body,head out of frame,
XL版本更新（Update）
The SDXL version optimizes the issues on the SD1.5 version
In SDXLv1 version, the trigger condition was changed to trigger the vine leaf effect under the low weight of this model and the high weight of other models without affecting other lora models as much as possible
The trigger word of the new version is changed to vines, which can be triggered normally under the weight of 0.6 to 0.8. When matching with other LORAs, the weight of other models' clothing needs to be reduced or the weight of the trigger word of this model needs to be increased When the weight is low, the weight of prompt words should be appropriately increased, including the natural language weight you want to trigger. The non-essential LORA weight should not exceed 0.8. In addition, the problem that SD1.5 affects body movements is solved Using natural language to trigger special cases can be combined with trigger words and natural language triggers
Such as
A girl with vines wrapped around her chest
a woman in a bodysuit with green leaves on her body and a hat on her head and hands on her head
a woman with green leaves on her body and a body paint job on her body and chest,
a woman with green leaves on her body and a green mask on her head and hands on her hips
a woman with green leaves on her body and a mirror in her hand
a woman with green leaves on her body and arms and legs
a woman with green leaves on her body and body paint on her body and hands and legs
a woman with green leaves on her body and stockings on her legs
More natural language can be found in my model metadata and I won't list them all
SDXL版本优化了SD1.5版本上的问题在SDXLv1版本中更改了触发条件可以在本模型低权重下其他模型高权重情况下触发藤蔓叶子效果而尽量不去影响其他lora模型新版本触发词改为vines 这个在0.6~0.8权重下都可以正常触发搭配其他lora的时候需要减少其他模型衣物的权重或者增加本模型触发词的权重 减少其他模型的衣服权重就用naked来触发就行 lora 低权重的时候需要适当的增加提示词的权重包括你想要触发的自然语言权重 非必要LORA权重不要超过0.8 另外还解决了SD1.5版本影响肢体动作问题现在可以自由控制肢体动作 由于SDXL 对自然语言理解比较好 完全可以不用触发词vines 可用自然语言来触发特殊情况可以配合触发词 和自然语言触发
例如
A girl with vines wrapped around her chest
a woman in a bodysuit with green leaves on her body and a hat on her head and hands on her head
a woman with green leaves on her body and a body paint job on her body and chest,
a woman with green leaves on her body and a green mask on her head and hands on her hips
a woman with green leaves on her body and a mirror in her hand
a woman with green leaves on her body and arms and legs
a woman with green leaves on her body and body paint on her body and hands and legs
a woman with green leaves on her body and stockings on her legs
更多的自然语言可以看我模型元数据里的我就不一一列举了
-----------------------------------------------------------------------------------------------------
1.1版本更新（Update）
1.1版本优化了此模型再搭配个别lora模型情况下会造成模型污染过拟合的现象
在1.1版本中更改了触发条件可以在本模型低权重下其他模型高权重情况下触发藤蔓叶子效果而尽量不去影响其他lora模型
新版本触发词改为(vines:1.5) 这个在0.6~0.8权重下都可以正常触发 不过由于权重越低影响其他lora就小 可在使用的时候从0.6权重开始逐步增加权重
Version 1.1 optimized this model and combined it with individual Lora models to cause model contamination and overfitting
In version 1.1, the triggering condition has been changed to trigger the vine leaf effect under low weight in this model and high weight in other models, while minimizing the impact on other Lora models
The trigger word in the new version has been changed to (vines: 1.5), which can be triggered normally at a weight of 0.6~0.8. However, the lower the weight, the smaller the impact on other Loras. When using it, gradually increase the weight from 0.6 to 0.8
其他没有提到的参数继续沿用上个版本
Continue to use the previous version for other parameters that were not mentioned
-----------------------------------------------------------------------------------------------------
1.1版本使用建议（ 1.1 version usage suggestions）：
推荐权重（Recommended weight）：0.6
新版本触发词改为(vines:1.5) 这个在0.6~0.8权重下都可以正常触发 不过由于权重越低影响其他lora就小 可在使用的时候从0.6权重开始逐步增加权重
The trigger word in the new version has been changed to (vines: 1.5), which can be triggered normally at a weight of 0.6~0.8. However, the lower the weight, the smaller the impact on other Loras. When using it, gradually increase the weight from 0.6 to 0.8
推荐负面提示词（Negative prompt words are recommended）：EasyNegative,(worst quality, low quality:2),monochrome,zombie,(interlocked fingers:1.2),extra limb,mutated hands and fingers,malformed hands,(head out of frame:1.5),headless,floating limbs,
-----------------------------------------------------------------------------------------------------
1.0版本使用建议（ 1.0 version usage suggestions）:
推荐权重（Recommended weight）：0.6~1.0 0.8为最佳 权重越低 身上叶子越少 默认可以不用加上触发词 只有在低权重下使用需要加上触发词
The recommended model weight is 0.6~1.0. 0.8 is the best. The lower the weight of your model, the less leaf effects it will have. In situations where the weight is high, triggering Lora effects can be added without adding triggering words
触发提示词（Trigger word）：vines,leaf,vine underwear,chest entangler vine
推荐负面提示词（Negative prompt words are recommended）：EasyNegative,(worst quality, low quality:2),monochrome,zombie,(interlocked fingers:1.2),extra limb,mutated hands and fingers,malformed hands,(head out of frame:1.2),headless,floating limbs,
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
- 模型反馈请站内私信
图像源和训练代码来自互联网，模型仅用于科研兴趣交流。如有侵权，请联系删除，谢谢。
The image source and training code are from the Internet, and the model is only used for scientific interest exchange. If there is infringement, please contact to delete, thank you.