Pumpkin Bat & Bat Girl

万圣节快乐！🎃 这是一个以万圣节为主题的Q版形象

Happy Halloween! 🎃 This is a Halloween themed Q version of the image

---

该 Lora 专注于生成具有良好构图的南瓜灯和萌化蝙蝠的可爱女孩，触发词为：bat \(animal\)、halloween、chibi

• This Lora focuses on making a bat cute girl with a well-composed jack-o'-lantern and a bat cute. The trigger words are bat \(animal\), halloween, chibi

本 Lora 使用 88 张图片进行训练，训练参数：epoch 为 20，每张图片学习 10 次，每两轮保存一次模型，基于 aniverse_thxEd14Pruned 底模训练。理论上使用该底模效果最佳，但经我测试，该 Lora 与其他底模的兼容性出乎意料地好。以下是我推荐的底模：

• This Lora was trained with 88 images. The training parameter epoch: 20, learn 10 times per image, save the model once every two rounds, based on aniverse_thxEd14Pruned base model for training. Theoretically, using this base model yields the best results, but after my testing, this Lora is surprisingly well-compatible with other base models. The following are the base models I recommend:

Movable figure model_v2.0 👈🏻点击前往模型主页 Click to the model home page

XXMix_9realistic_v4.0 👈🏻点击前往模型主页 Click to the model home page

---

虽然南瓜和蝙蝠在 SD1.5 基础模型中已存在一部分，因此您无需 Lora 即可生成它们，但若您尝试在南瓜和蝙蝠中加入其他元素，生成结果可能会出现不协调或奇怪的情况。建议您在使用本模型前，对比测试使用与不使用我的 Lora 的效果差异。相信我，您会有所发现。

Although pumpkins and bats are part of the SD1.5 base model, so you don't need Lora to make them, if you try to include something in pumpkins and bats, the resultant combination may become disobedient or weird. You can test the difference between using my Lora and not using my Lora before you use my model. Trust me, you'll find something new.

---

推荐参数（Recommended parameter）：

推荐封面图起始 TAG：bat \(animal\), halloween, 3D graphics, chibi

采样方法（sampler）：

euler a：20~30 步

dpm++ 2s a karras：25~30 步

dpm++ sde karras：20~30 步

dpm++ 2m sde karras：20~40 步

dpm++ 3m sde karras：45 步及以上

ddim：30 步及以上

提示词引导系数（cfg scale）：3-7

人物分辨率（character resolution）：512x768、768x768、512x1024

高清修复（hires.fix）：

放大算法：r-esrgan 4x+anime6b、4x-ultrasharp、4x_smolface_200k

高分辨率迭代步数：10~15

重绘幅度：0.3~0.5

clip skip：2

负面提示词推荐（negative）：

推荐一组负面提示：(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, text, logo, username, multiple moles, mole on body:1.2)

推荐使用嵌入模型中的 Easynegative 或 Easynegativev2

可单独使用负嵌入模型出图

---

模型用途声明

1. 您不得将本模型及其衍生版本（如融合模型版本）托管于计划盈利或接受捐赠的网站/应用程序。若您需将本模型及其衍生版本用于商业目的（如生成式服务、售卖图像、将图像用于短视频平台、文章或出版物等公开传播），请通过站内私信联系本人。

2. 您不得直接售卖本模型生成的图像，除非您对该图像进行了充分的人工修改，使其在法律意义上可被完全认定为您的原创作品。若违反本条款，由此产生的一切法律后果由您自行承担，本人概不负责。

3. 您不得使用本模型故意生成或传播非法、有害内容。请遵守公序良俗，将本模型用于积极正面的用途。

Model purpose statement

1. You may not host this model and its derivative versions (such as the Fusion Model version) on websites/applications that plan to generate income or donations. If you need to use this model and its derivative versions for commercial purposes (such as generating services, selling images, using images for public publication in articles or publications on short video platforms, etc.), please contact me via the site.

2. You may not directly sell images generated from this model unless you have manually modified the image to a sufficient degree that it can be legally judged to be your own work. If you violate this article, all legal consequences caused by you personally bear, I will not be responsible.

3. You cannot use this model to intentionally create or share illegal or harmful content dissemination and output. Please abide by public order and good customs, and use this model for positive purposes.

• 模型反馈请通过站内私信

本模型不包含任何真人素材图像，训练代码来自互联网，仅用于科研兴趣交流。

This model does not contain any real human material image source and training code from the Internet, the model is only used for scientific research interest communication.