烟上月|写实摄影_十指流玉（Realistic photography）

Introduction in Chinese：

这是一个致力于写实风格的SDXL人像微调大模型，旨在改善SDXL对于国人形象的不足之处，大幅减弱了了原版SDXL人像里的日韩味儿、欧美味儿，替换成了纯正的中国人面貌形象。本模型（以下简称十指流玉）尤为擅长刻画写实人像，同时也能生成各种写实类型的背景、角色IP形象、动植物等等一切基于现实存在的事物。使用十指流玉当底模训练出来的lora，出图的时候能获得更好的国人形象。十指流玉主要使用了三千多张基于现实的胶片摄影集进行训练，泛化性非常好，兼容几乎所有的写实类lora。更适合中国宝宝体质的国人大模型！我在训练素材中加入了一部分幻想题材的原画集，使得该模型极具想象力，能描绘出各种超现实的背景，非常适合大家拿去做AI摄影，替换实拍照片的背景，换成各种幻想类的奇特背景（炼丹交流答疑Q群：892762841，欢迎大家前来交流心得，我会经常为大家释疑解惑）

十指流玉不干扰任何脸型的生成！切换提示词生成的每一张脸几乎都是不同的脸型。这意味着你可以轻松使用Adetailer插件对图片进行指定人脸的更换生成，以及使用任意脸型lora改变出图的脸型特征。（如果你不了解这个插件，可以去B站搜一下ADetailer，学一下使用方法）也可以直接随机抽卡，抽出你的SSR哦~完全不必担心每张脸都是固定的

喜欢十指流玉的话，可以给我的模型点个爱心并给个五星评价哦，这对我是莫大的支持！非常感谢♪(･ω･)ﾉ 如果有生成好看的图片，可以给我返个图~

关于采样方法，不做要求，我测试过几乎都能用，但我强烈推荐Restart！这个优化器是新出的算法，只需10-15步就能有非常好的出图效果，潜力极大，且兼顾了速度和质量。我的封面也基本都是用的这个采样方法。说个题外话，很多采样方法每次都会在出图过程中加入新的随机噪声，导致图片无法复现，这是很正常的现象。想要百分百复现原图，需要使用euler a出图

关于提示词的写法，不需要写一大堆提高质量的tag！比如什么高质量图片，8k，极其详细的细节，精美图片，超级精细的插画，这些提示词在十指流玉模型里完全没必要添加。不加这些提高质量的提示词，也可以生成高精度图片。负面提示词也是一样的道理，比如什么畸形的手，多余的脚，这些都不需要添加！！！它们并不会对图片的质量带来影响。大家写提示词只需要聚焦于你想描述的画面场景即可，十指流玉对于自然语言的提示词也能很好的理解出图，可以多多尝试用完整的语句去描述你想刻画的画面结构。

图片分辨率没啥要求，只需注意最低分辨率尽量不要低于1024即可，1024以上的分辨率才是十指流玉发挥写实人像优势的最低要求

有需要定制模型、商业合作、一对一教学的，可以加我Q：928075483。

抖音小红书同名——烟上月，账号会经常发布一些漂亮的赛博cos小姐姐，感兴趣可以去关注下~

喜欢我的作品可以爱发电支持一下我哦，爱发电会定期发布精心制作的独占模型，以及往期作品的迭代改进版本（爱发电直接搜 烟上月，就能找到我）

最后欢迎各位提出优化建议和改进方向

————————————————————————————————————————

模型用途声明 1. 您不得将此模型及其衍生版本（如融合模型版本）托管于计划赚取收入或捐赠的网站/应用程序。如果您需要将此模型及其衍生版本用于商业目的（如生成式服务、售卖图片、将图片用于公开发表的文章或出版物等），请通过Q联系本人，928075483

2. 您不得直接售卖由此模型生成的图片，除非您对该图片进行了足够程度的人工修改，使其在法律意义上可以被完全判定为您的个人作品。如果您违反本条，所造成的一切法律后果由您个人承担，请恕本人概不负责。

3. 您不能使用该模型故意制作或共享非法或有害的输出或内容，请您遵守公序良德，将此模型用于积极正面的用途。

English introduction：

This is a SDXL portrait fine tuning large model based on film photography material training, this model (hereinafter referred to as ten finger Liuyu) is particularly good at portraying real portraits, but also can generate a variety of realistic types of background, role IP image, animals and plants and so on all things based on real existence. The lora trained by using ten fingers flowing jade as the bottom mold can get a better image of the Chinese people when drawing. Ten Fingers Liuyu mainly uses more than 3,000 reality-based film photography albums for training, generalization is very good, compatible with almost all realist LORA

Ten fingers flow jade does not interfere with the formation of any face shape! Almost every face generated by switching prompt words is a different face shape. This means that you can easily use the Adetailer plugin to change the face of the image, and use any face shape lora to change the face shape of the image. (If you don't understand this plug-in, you can go to Station B to search ADetailer and learn how to use it) You can also draw the card directly at random and pull out your SSR Oh, you don't have to worry about each face is fixed

If you like ten fingers, you can give my model a love and give a five-star evaluation oh, this is a great support for me! Thank you very much. ♪(· ω ·)ノ if there is a good picture generated, can send me a picture back ~

As for the sampling method, there are no requirements, I have tested almost all work, but I highly recommend Restart! This optimizer is a new algorithm, only 10-15 steps can have very good results, great potential, and both speed and quality. I use this sampling method for most of my covers. As a side note, many sampling methods will add new random noise to the drawing process each time, resulting in the image will not be reproduced, which is a normal phenomenon. To reproduce the original drawing 100%, use euler a to draw

As for how to write prompt words, there is no need to write a lot of quality improving tags! For example, what high-quality pictures, 8k, extremely detailed details, beautiful pictures, super detailed illustrations, these prompt words are completely unnecessary to add in the ten-finger Liuyu model. It is possible to produce high-precision images without these quality enhancing prompts. The same goes for negative cues, such as deformed hands or extra feet, which don't need to be added!!! They do not affect the quality of the image. We write prompt words only need to focus on the scene you want to describe, ten fingers Liuyu for natural language prompt words can also be a good understanding of the figure, you can try to use a complete sentence to describe the structure of the picture you want to depict.

Picture resolution is not required, just pay attention to the minimum resolution as far as possible not less than 1024, 1024 resolution is the minimum requirement for ten fingers Liuyu to play realistic portrait advantages

If you need customized models, business cooperation, one-to-one teaching, you can add me Q: 928075483. Tiktok little red book with the same name - Smoke last month, the account will often release some beautiful cyber cos little sister, interested can go to pay attention to ~ C station name: Yan_shangyue, will release some NSFW content, https://civitai.com/user/Yan_shangyue/models like my works can love power support oh, my love power regularly publish elaborate monopoly model, And the iterative and improved version of the previous work (Love power generation directly search smoke last month, you can find me) Finally welcome you to propose optimization suggestions and improvement directions

-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --

1. You may not host this model and its derivative versions (such as the Fusion Model version) on websites/applications that plan to earn income or donate. If you need to use this model and its derivative versions for commercial purposes (e.g. generative services, sale of images, use of images in published articles or publications, etc.), please contact me at Q, 928075483

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 the model to intentionally create or share illegal or harmful output or content. Please abide by the public order and use the model for positive purposes.