Flux高清人物Lora模型训练采用超过50张高清图片训练而成，训练总步数超过10000步。本模型为V1.5版本，是V1.0的减半精度版本。如果觉得V1.5版本不能满足需求，可以去我的P站主页下载更高精度的版本。现阶段主要推出两类Lora模型一类是以DIM值64的模型V1.5，一类是以DIM值128的模型V1.0，两者在实际实用过程中，效果差异主要在纹理细节不同，光影不同等方面。支持我，将会有更多更好的Lora炼制出来，感谢！My Patreon Site<a target="_blank" rel="ugc" href="https://www.patreon.com/c/Fairytale2399">https://www.patreon.com/c/Fairytale2399</a>如需定制Lora，可在我的Pateon页面上留言。If you need to customize Lora, you can leave a message on my Pateon page.Flux HD Character Lora ModelThe training is conducted using over 50 high-definition images, with a total of over 10000 steps.This model is version 1.5, which is a half precision version of version 1.0.If you feel that version 1.5 cannot meet your requirements, you can download a higher precision version from my Patreon website homepage.At present, two types of Lora models are mainly introducedOne is the model V1.5 with a DIM value of 64, and the other is the model V1.0 with a DIM value of 128. In practical applications, the differences between the two mainly lie in texture details, lighting, and other aspects.Support me, there will be more and better Lora refining, thank you!声明： 本模型内容仅为知识分享与学术交流目的，模型的使用均基于合法合规前提展开，禁止用于任何违反所在国家/地区法律法规的用途。Disclaimer: The content of this model is only for the purpose of knowledge sharing and academic exchange. The use of the model is based on legal and compliant premises, and is prohibited from any use that violates the laws and regulations of the country/region where it is located.

jiguang

Can be used without triggering words, it is recommended to add Asian faces for better controllability

Files with SHA256: 4bf1396fd0f45e434b4d72d3c2c4043a9b1fa023f8d6466e92fbf5dd3db1e76c