******************* 关于版本及训练集 *******************

----> 目前是V1.0版本，也是我自己的一个尝试版本，泛化性上有待优化，有时间会再更新优化版本

----> 训练底模是 helloyoung25d_V10f

----> 触发词：guangyu/sky kid (非必需)

----> 训练集图片来自游戏原始截图和部分ps处理过的游戏截图

----> 训练集：589张 | 步数：39540 | 训练器：lora-scripts-v1.8.1

【免责声明】

1. 原则上本模型仅用于交流学习，不得将模型及生成图用于商业用途。

2. 使用者在遵守本国法律法规和本服务条款的条件下使用，基于本模型生成的内容产生的任何知识产权纠纷问题由使用者自行处理。

3. 未经作者本人授权同意，不得将本模型服务集成在任何第三方软件或服务中。

******************* About version and training set *******************

----> It is currently the V1.0 version, which is also a trial version of my own. The generalization needs to be optimized. I will update the optimized version when I have time.

---->The training model is helloyoung25d_V10f

----> Trigger word: guangyu/sky kid (optional)

----> The training set pictures come from the original screenshots of the game and some PS-processed game screenshots.

----> Training set: 589 images | Number of steps: 39540 | Trainer: lora-scripts-v1.8.1

【Disclaimer】

1. In principle, this model is only used for communication and learning, and the model and generated diagrams are not allowed to be used for commercial purposes.

2. Users shall use it under the conditions of complying with the laws and regulations of their country and these terms of service. Any intellectual property disputes arising from the content generated based on this model shall be handled by the user themselves.

3. This model service may not be integrated into any third-party software or service without the authorization and consent of the author.

******************* 使用建议 *******************

【底模使用建议】

----> 建议使用动漫、游戏、插画类的，人物面部更扁平的模型更容易出效果，不适合真实系模型。

----> 适配度较好的模型有：

hello系列非真实系大模型、动漫/25D大模型（revAnimated系列）、插画类大模型、可爱风模型

* 以上底模为部分推荐，也可以根据自己的喜好尝试~

【Lora权重建议】

----> 若使用二次元和插画类的底模，Lora权重在0.5-0.8更好，若使用2.5D类的，Lora权重在0.6-0.9更好。如果只想生成纯场景，加入map/scene/venue/environment/background等提示词或提高某地图名称关键词权重。

*（因为角色的面部特征比较特殊，太高太低容易坏脸。如果不想要面具可以搭配局部重绘修改）

参数设置建议：

图片尺寸：768*768、1024*1024、768*1024、900*1200等（不建议过长或过宽的比例）

采样方法：Euler a, DPM++ 2M Karras,

采样步数：24-30

剪辑跳过：2

重绘幅度：0.2-0.5

CFG比例：7

VAE ：看情况使用

******************* Recommendations *******************

[Suggestions for using the bottom mold]

----> It is recommended to use animation, games, and illustration models. Models with flatter characters' faces are more likely to produce results and are not suitable for realistic models.

----> Models with better fitness include:

hello series non-realistic large models, animation/25D large models (revAnimated series), illustration large models, cute style models

*The above base models are only some recommendations, you can also try them according to your own preferences~

[Lora weight suggestion]

----> If you use two-dimensional and illustration type base models, the Lora weight is better between 0.5-0.8. If you use 2.5D type, the Lora weight is better between 0.6-0.9. If you only want to generate pure scenes, add prompt words such as map/scene/venue/environment/background or increase the keyword weight of a certain map name.

*(Because the character’s facial features are quite special, if it is too high or too low, it will easily damage the face. If you don’t want a mask, you can modify it with partial redrawing)

Parameter setting suggestions:

Image size: 768*768, 1024*1024, 768*1024, 900*1200, etc. (too long or too wide ratio is not recommended)

Sampling method: Euler a, DPM++ 2M Karras,

Number of sampling steps: 24-30

Clip skipped: 2

Redraw range: 0.2-0.5

CFG ratio: 7

VAE: Use according to the situation

******************* 提示词使用说明书 *******************

----> 在使用人物提示词的时候，如果想要更贴和游戏内的样子（斗篷和发型动作等）可以用1 sky kid或sky kids，如果是1boy或1girl，人物表现会更灵活点。

----> 目前版本包含部分发型，斗篷裤子、地图和部分动作，具体如下：

1、地图：

云野：yunye,butterfly plain,floating island

云峰：yunfeng,cloud peaks

雨林：yulin,rainforest

霞谷：xiagu,Valley of Triumph,Xiaguang City

云巢：yunchao,Cloud Nest,small town

暮土：mutu,Golden Wasteland

2、斗篷：各色一级斗篷（直接用颜色词加cape），圣诞斗篷Christmas cape，白绒斗white velvet cape，彩色围巾斗篷colorful scarf cape，粉色花瓣斗pink petals cape，彩虹斗rainbow cape，护林员斗篷ranger cape，破旧斗篷worn cape

3、裤子：短裤shorts，武士裤samurai pants，拖地长裤floor-length trousers，长裙dress，音韵季裤子Rhyme Season Pants，风行季蓝色裤子blue pants for the Season of Flight，风行季橙色裤子orange pants for the Season of Flight

4、发型：初始短发initial short hair，马尾辫ponytail，龙骨keel,keel hairstyle，花环wreath，雨林长发rainforest long hair，正太Shota，猫耳cat ear，卡卡kaka，菇菇gugu，刘海短发short hair with bangs，双牛角辫double horn braid，丸子头puff ball hairstyle，箬笠Ruo Li，绿丝带green ribbon，明制帽Ming hat，女巫发型witch hairstyle，双麻花辫double twist braids，齐肩长发shoulder-length hair，脏辫短发dreadlock short hair

5、动作：站立stand，引蝶attract butterflies/catch butterflies，抱拳clasping fists salute，拥抱hug，牵手hold hands，飞吻blow a kiss，跳舞dance

******************* Prompt Word Instructions *******************

----> When using character prompt words, if you want to be more consistent with the appearance in the game (capes, hairstyle, actions, etc.), you can use 1 sky kid or sky kids. If it is 1 boy or 1 girl, the character performance will be more flexible. point.

---->The current version includes some hairstyles, cloaks and pants, maps and some actions, as follows:

1. Map:

Daylight Prairie/BIRD NEST：yunye,butterfly plain,floating island

PRAIRIE PEAKS：yunfeng,cloud peaks

Hidden Forest：yulin,rainforest

Valley of Triumph：xiagu,Valley of Triumph,Xiaguang City

Town:yunchao,Cloud Nest,small town

Golden Wasteland：mutu,Golden Wasteland

2. Capes: first-level capes of various colors (directly add cape to the color word), Christmas cape, white velvet cape, colorful scarf cape, pink petals cape, rainbow cape, forest ranger ranger cape, worn cape worn cape

3. Pants: shorts, samurai pants, floor-length trousers, long skirts, Rhyme Season Pants, blue pants for the Season of Flight, orange pants for the Season of Flight pants for the Season of Flight

4. Hair style: initial short hair, ponytail, keel, keel hairstyle, wreath, rainforest long hair, shota, cat ear, kaka, mushroom gugu, short hair with bangs with bangs, double horn braid, puff ball hairstyle, Ruo Li, green ribbon, Ming hat, witch hairstyle, double twist braids, shoulder-length hair hair,dreadlock short hair

5. Actions: stand, attract butterflies/catch butterflies, clasping fists salute, hug, hold hands, blow a kiss, dance

【免责声明】

1. 原则上本模型仅用于交流学习，不得将模型及生成图用于商业用途。

2. 使用者在遵守本国法律法规和本服务条款的条件下使用，基于本模型生成的内容产生的任何知识产权纠纷问题由使用者自行处理。

3. 未经作者本人授权同意，不得将本模型服务集成在任何第三方软件或服务中。

【Disclaimer】

1. In principle, this model is only used for communication and learning, and the model and generated diagrams are not allowed to be used for commercial purposes.

2. Users shall use it under the conditions of complying with the laws and regulations of their country and these terms of service. Any intellectual property disputes arising from the content generated based on this model shall be handled by the user themselves.

3. This model service may not be integrated into any third-party software or service without the authorization and consent of the author.